THE MAGAZINE FOR COMPUTING

& DIGITAL MAKING EDUCATORS

ARTIFICIAL INTELLIGENCE
What does AI mean for computing education?

BIAS IN THE MACHINE

How to address gender balance in AI

NO ROBOTS
How anthropomorphism hinders AI education

Issue 22 Oct 2023 helloworld.cc

PLUS
3D WORLDS IN UNITY • ONLINE COURSES FOR PROFESSIONAL GROWTH • TEACH FIRST • T LEVELS
1 (helloworld.cc)
START A CODE CLUB • SEAME FRAMEWORK • BBC MICRO:BIT • WIPSCE 2023 TAKEAWAYS
ASTRO PI • INTRO TO AI LESSON • GREEN SCREEN IN THE CLASSROOM • SYSTEMS THINKING
Teach and learn with the
Raspberry Pi Foundation
Free for everyone anywhere in the world

Teachers’ resources Project library

Discover training, resources, and Browse 200+ online project guides
guidance to help you teach computing that help young people create with
with confidence. digital technologies.

raspberrypi.org/teach projects.raspberrypi.org

Research to practice Coding at home

Explore our research into what works Watch our support tutorials and access
best in teaching school-age young engaging resources for your child.
people about computing.
raspberrypi.org/learn
raspberrypi.org/research

Raspberry Pi Foundation, UK registered charity 1129409

 Hello World is the
official magazine
of the Raspberry
Pi Foundation

EDITORIAL
HELLO, WORLD! IBE
Editors
SUBSCR Jan Ander, Gemma Coleman,
IN PRINT
FORRN TFORPAEGEE
Ben Garside, Meg Wang

E
ven though it’s been less than a year
Subeditors
since the release of AI applications TU Louise Richmond and Amy Rutter
such as OpenAI’s ChatGPT and 22 Subscriptions
Google’s Bard, it feels as though they’ve been Sean Sayers
around for a long time. Certainly their release Social Media
has made people pay attention to what AI It’s inspiring for me to see how the education Neena Patel
technology is, and could be, capable of. community is reacting to the opportunities that
When I was in the classroom, I used to enjoy AI can provide. Of course, there are elements DESIGN
getting to grips with new technological of AI where we need to tread carefully and be
criticalmedia.co.uk
advances and finding ways in which I could very cautious in our approach, but what you’ll
Head of Design
bring them into school and excite the students I see in this magazine is educators who are
Lee Allen
taught. Occasionally, during the busiest of thinking creatively in this space. Sue Sentance
Designer
times, I’d also look longingly at other subjects (page 28) highlights how AI applications can Sam Ribbits
and be jealous that their curriculum appeared to reinforce gender bias and stereotypes; Sway Photography
be more static than ours (probably a huge Grantham (page 32) offers tips for how to talk The Raspberry Pi Foundation,
misconception on my behalf). What I think is to young children about AI; and Ken Kahn Adobe Stock

really exciting about developments in AI is the (page 54) explores how chatbots could affect Graphics
Rob Jervis
impact they are having on education as a the learning of
Cover
whole, and this most definitely isn’t just the programming.
© MUTI
remit of computing. Philip Colligan talks in his We hope this

CONTRIBUTORS
article on page 23 about viewing AI education issue will provide
as three distinct areas: you with plenty of Edward Abel, James Abela, Sophie Ashford, Pete
Bell, Halima Bhayat, Mac Bowley, Mark Calleja,
n AI literacy: what it is and how we teach it ideas to take away
Katharine Childs, Gemma Coleman, Philip Colligan,
n Rethinking computer science (and possibly and build upon. Michael Conterio, Veronica Cucuiat, Catherine Elliott,
Tracy Gardner, Ben Garside, Sway Grantham, Ben
some other subjects) Hall, Claire Johnson, Ken Kahn, Gitika Kaur, Fergus
Kirkpatrick, Dale Lane, Rebecca Mack, Tessy Mosindi,
n Enhancing teaching and learning through Ben Garside
Rujeko Moyo, Alan O’Donohoe, Richard Pawson,
AI-powered technologies Guest Editor Judy Robertson, Chris Roffey, Sue Sentance, Daljit
Shoker, Jane Waite, Aimee Welch, Magda Wood,
Seema Zerafa

Supported by
NICE TO MEET YOU
Contributing Partner

MEG WANG
My name is Meg, and I’m your new Editor. I’m thrilled to take up the task of connecting This magazine is printed on paper sourced from
this wonderful community of educators, volunteers, parents, and researchers through our
sustainable forests and the printer operates an
environmental management system which has
print and digital magazine. While I’ve worked at the Raspberry Pi Foundation for a couple been assessed as conforming to ISO 14001.

of years, prior to joining, I had a long career in book publishing, and I am still learning Hello World is published by the Raspberry Pi
Foundation, 37 Hills Road, Cambridge, CB2
about computing, digital making, and educational practices. My passion for computing 1NT. The publisher, editor, and contributors
accept no responsibility in respect of any
education comes from what young people can do with the knowledge and skills they omissions or errors relating to skills, products,
or services referred to in the magazine.
gain in the classroom. I’m always thankful to Hello World contributors and experts like Except where otherwise noted, content in this
magazine is licensed under a Creative Commons
this issue’s guest editor, Ben Garside, who share their insights, ideas, and knowledge in Attribution-NonCommercial-ShareAlike 4.0
a way that’s accessible and practical for people like me who are newer to the subject. Unported (CC BY-NC-SA 4.0).

If you have an idea for an article or just want to say hi, drop me a line (contact@
helloworld.cc). I look forward to learning from you and alongside you.

helloworld.cc 3
 CONTENTS
23–57 COVER FEATURES

© Monkey Business/stock.adobe.com
TEACHING AND AI
A series of articles exploring teaching
about artificial intelligence

NEWS, FEATURES, AND OPINION

6 NEWS 18 I NTERVIEW 30 EXPERIENCE AI
Research centre; micro:bit — the Dale Lane, founder of Machine A new education programme
next gen; Unity projects; European Learning for Kids, discusses making AI accessible and relevant
Astro Pi Challenge; meet St Joseph’s AI tools in education
Secondary School in Rush, Ireland 32 NOT ROBOTS
How to talk to young children
INCLUSION COLUMN 22 SUBSCRIBE TODAY!
12 about AI
Artificial intelligence and the inclusive
computing classroom 34 CHILDREN AND SMART
23 A I AND COMPUTING EDUCATION SPEAKERS
14 T HE SEAME FRAMEWORK Breaking down what AI can mean for Research exploring children’s
A useful tool for teaching AI and computing education understanding of AI
machine learning lessons
26 A I AND ANTHROPOMORPHISM 36 AI FOR 9–13-YEAR-OLDS
16 I NSIGHTS — AI ETHICS How human-based descriptions feed AI lesson ideas for primary
The SUM values our misconceptions about AI and secondary students

17 I NSIGHTS — PRODABI 28 B
 IAS IN THE MACHINE 38 P ICTURING AI
Human–machine ecologies Ways to address gender bias in AI Exploring better images of AI

4 helloworld.cc
 LEARNING
RESOURCES & LESSON PLANS

CODE CLUB
Four groups share
their experiences of
starting a Code Club 58

40
© Teresa Berndtsson/CC-BY 4.0

AI TOOLS FOR LEARNING

Programming with ChatGPT 70 INFORMATION TECHNOLOGY
IT all around us

72 INTRO TO AI
A lesson from Experience AI

CONVERSATION
© ImageFlow/stock.adobe.com

EXPLAINING AI
54 How we explain AI
terms to young people 75
BEBRAS
A fun challenge in
computational thinking

GREEN SCREEN
76
40 EXPLAINING AI 56 K
 EEP PROGRAMMING Developing learners’ digital skills with
How we created a glossary of AI Learners still need hands-on a movie-making competition
terms for young people programming practice
INSIDER’S GUIDE
42 MODEL CARDS 58 C
 ODE CLUB 78 Work-related learning experience
Teaching AI explainability Four groups share their experiences
of starting a Code Club
44 RECOMMENDER SYSTEMS T LEVELS
82
The hidden AI steering our choices
60 C
 OMPUTER SCIENCE AND Tips from one of the first T-level
GEOMETRY pilot schools
46 DEMYSTIFYING MACHINE Linking computer science
LEARNING EVOLVING COMPUTING
Activities to unpack AI in
and maths concepts through 84 Teaching systems thinking
computational thinking
recommendation systems

TIKTOK’S ALGORITHMS 62 O
 NLINE COURSES
48 Tips for making the most of online
Can AI and data create learning for CPD

REVIEWS
more inclusion?

AI DEBATE
64 T EACH FIRST
50 Trainee teachers share how they put
Do the benefits of AI outweigh computing pedagogies into practice
the risks?
85 BOOK REVIEW
67 W
 IPSCE 2023
54 AI TOOLS FOR LEARNING
Teach Data Literacy
Teacher thoughts from a computing
Programming with ChatGPT education research conference

helloworld.cc 5
NEWS FEATURE

A NEW INITIATIVE
FOR RESEARCH IN
COMPUTING EDUCATION
The Raspberry Pi Computing Education Research Centre is working collaboratively with teachers
to understand more about the teaching and learning of computer science in schools

Sue Sentance

I
n issue 19 of Hello World, we and justice. It has been used in computing to the same survey globally in spring 2024, so
announced the launch of the develop engaging and responsive computing watch out for updates! This work helps us to
Raspberry Pi Computing Education teaching for all students, regardless of their compare the experiences of computing
Research Centre, a joint initiative between social or ethnic backgrounds. In 2022, we teachers around the world, and makes an
the University of Cambridge and the ran a project in schools to investigate how important contribution to policy, as well as
Raspberry Pi Foundation teachers were implementing culturally providing general information for teachers.
(computingeducationresearch.org). The responsive computing teaching. A
mission of the centre is to carry out technical report described our work, and Conferences and events
rigorous and impactful research around in August we also presented a research We hosted the WiPSCE conference at the
the teaching and learning of computing, paper about the first part of our findings end of September (wipsce.org) and are
with a particular focus on those who (helloworld.cc/CRPresearch). planning to host or run other research
might not have ready access to computing. events in the future. We also present our
So, what have we been doing since our The Google DeepMind school work at other conferences: for example,
official launch in 2021? engagement programme Laurie Gale presented a paper on
The university has been contracted to debugging at the UKICER conference in
PhD students be the independent evaluator of the six September (ukicer.com), and I presented a
A key function of the centre is to train PhD programmes that Google DeepMind is paper on culturally responsive computing
students who will work on projects relating running as part of its school engagement teaching (helloworld.cc/CRTpaper) at the
to its core work. We are delighted to be initiative to broaden participation in AI and ICER conference in August.
supporting two new PhD students, Laurie STEM. This project will run until 2024 and
Gale and Salomey Afua Addo. Laurie is we will be reporting our findings to Google Teacher research network
focusing on debugging strategies in DeepMind, as well as publishing specific We are keen to collaborate with teachers
secondary-school programming, and insights (helloworld.cc/nextgenAIleaders). in our research, and so we are in the
Salomey, who is partially supported by the process of creating a teacher research
Micro:bit Educational Foundation, is Surveying computing teachers network in computing. This will aim to
investigating AI education in school, which In 2022 we surveyed teachers across the UK facilitate classroom-based research, as
will include fieldwork in Ghana. and Ireland, and published a paper in which well as keeping teachers in touch with
we compared computing education policy research developments.
Culturally relevant pedagogy and practice in Ireland and the countries of Find out more about all these initiatives
and computing the UK. Since then, we’ve published a and other research projects from the
Culturally relevant pedagogy is a teaching summary of the responses to the survey, and Raspberry Pi Foundation at
approach that has been developed to a second paper will soon be published computingeducationresearch.org, where
address underrepresentation in certain areas around teachers’ experiences of CPD you can also sign up for our newsletter and
of education, and focuses on social equity (helloworld.cc/UKICTS). Our plan is to run watch a video describing our work.

6 helloworld.cc
 NEWS FEATURE

BBC MICRO:BIT -
THE NEXT GEN
Get free micro:bit devices, resources, and training
for your primary school in the UK

Magda Wood, Chief of Learning, Micro:bit Educational Foundation

B
BC micro:bit – the next gen is a primary schoolchildren will end up working Micro:bit Educational Foundation’s three-
campaign that aims to increase in jobs that do not exist today (helloworld.cc/ phase national programme and primary
teacher knowledge and confidence wef2016). How can we best prepare a child pilot project (microbit.org/impact/research).
through the provision of micro:bit devices, for a future career that does not yet exist? This was carried out in collaboration with
lesson plans, and supporting materials. In such a landscape, it is unsurprising Nominet with the aim of better
From this autumn, 700,000 free micro:bit that teachers feel overwhelmingly understanding the difficulties faced by
devices (up to 30 per school) will be made unprepared when it comes to teaching primary teachers in the UK, and to enhance
available to primary schools throughout digital skills. Research carried out by the the teaching of digital skills in primary
the UK, as well as virtual training to Micro:bit Educational Foundation and classrooms. We look forward to seeing how
support teachers and help them make an Nominet last year indicated that 61 percent the programme can empower teachers and
impact in the classroom. of UK primary teachers responsible for inspire millions of young people to live their
teaching computing have no background in best digital lives.
Enhancing digital education the subject, with three in five teachers also If you are a primary teacher from
Computing and digital literacy are fast citing a lack of resources as a major barrier. anywhere in the UK, you can register at bbc.
becoming foundational skills — not just to The BBC micro:bit – the next gen project co.uk/teach/microbit. Registration closes on
reduce the digital skills gap the economy is aims to equip and support schools with the 18 December 2023. Registered teachers
already struggling with, but also to ensure right tools and resources to foster should expect their packs by March 2024 at
that young people have the skills to work computational thinking, inspire digital the latest.
the increasingly technical jobs of the future. creativity, and enhance the curriculum with
Research suggests that 65 percent of innovative machine learning lessons.

Real-world applications
ABOUT THE CAMPAIGN
© Courtesy of the BBC

School children have already used the

BBC micro:bit – the next gen will be made
micro:bit to tackle real-world problems such
possible by the following partners:
as developing flood early-warning systems
to notify coastal areas of bad weather and
• The free micro:bits will be funded by the
developing radio messaging systems to
UK registry Nominet
boost communication in the classroom.
New resource packs will launch as part • Teacher training and education resources
of BBC micro:bit – the next gen. These will be provided for free by the Micro:bit
include lessons to help teachers get started Educational Foundation
with physical computing, and engaging
projects to introduce working with the • The capabilities of the micro:bit will be
micro:bit in science investigations, data showcased with the help of favourite
logging, and working with digital data and CBBC and family entertainment brands
machine learning. from BBC Education, to demonstrate how
The next mission of the BBC micro:bit – simple and accessible it is for teachers to
the next gen project is to assist all students use in the classroom
n Two primary schoolchildren using the micro:bit device and teachers in the UK, building on the

helloworld.cc 7
NEWS FEATURE

UNLEASHING CREATIVITY:
3D INTERACTIVE WORLDS
IN UNITY
The Raspberry Pi Foundation’s Unity projects can spark a passion for
3D-game design and programming through new challenges

Pete Bell

3
D interactive experiences have pathways in which creators explore and which creators explore, design, and finally
become increasingly prevalent in create 3D worlds. Unity is a cross-platform, invent their own projects. The first pathway,
our lives. From captivating video games, to real-time, 3D-development platform that ‘Introduction to Unity’, includes step-by-
stunning animations, to immersive virtual has been used to create some super- step guides to using 3D objects, character
reality experiences, the world is evolving popular games such as Subway Surfers and controllers, colliders, text, and buttons to
into a realm where imagination knows Among Us. This can be really motivating create 3D worlds with characters. Starting
no bounds. Highlighting the ubiquity and for young creators embarking on these new off at the Explore stage, learners make a
appeal of these experiences can help project paths. The ability to deploy their mini game collecting sparkling stars, and
educators to capture the attention and creations across mobile, desktop, and web design non-playable characters (NPCs)
interest of young creators. offers them the opportunity to showcase for interaction. These projects build a solid
Teachers will have seen that many their creations to a broader audience and foundation of skills and confidence, and
young creators are inspired to begin their widens their scope as developers. familiarise learners with Unity’s interface,
programming journey because they play scripting, and asset management.
games and want to develop something Introduction to Unity Creators are then ready to move on to the
similar. To help young people find a place The two project pathways are both Design stage, which allow learners to take
to start, the Raspberry Pi Foundation has structured around our ‘3 … 2 … 1 … Make!’ the reins and build captivating worlds with
partnered with Unity to develop two project approach to project-based learning, in unique features that incorporate NPCs and

n The 3 … 2 … 1 … Make! approach to projects is informed by research into how young people develop interest and skills in computing

8 helloworld.cc
 through a collapsing tile floor ... with added
fireworks if you get through it, obviously!
Creators next move onto the Design
stage of the project, where they use their
newly developed skills to bring their inner
artist to life. They develop unique pixel art
on a tiled floor and reveal their awesome
artwork by rolling a ball across the surface.
Soon they’re stepping into the shoes of a
game developer by designing custom tracks
filled with obstacles, personalised effects,
sounds, and more!
The pathway culminates with the Invent
stage, aptly named ‘Marble Mayhem’.
Learners can bend the laws of physics to
create a chaotic one-of-a-kind game.

Powerful programming, individualised

n Crossy Rat, an example game created from one of the pathways interactivity
Young creators who have shown interest in
quests. This stage promotes critical ‘Introduction to Unity’ pathway, but we have Scratch animations or Python and want to
thinking and problem-solving as creators been careful to include reminders of the take the next step can try all of these projects
tackle design challenges and implement key skills required throughout the projects. out in C#. This is a widely used, versatile,
their ideas. Starting off at the Explore stage, creators object-oriented programming language that
Finally, learners reach the Invent stage, are introduced to the pathway theme of provides students with transferable skills that
and are encouraged to create something marble puzzles. Learners use premade can be applied beyond game development.
entirely unique. With the support of the assets to create a 3D-simulation adventure This expands learners’ programming
skills developed throughout the pathway, along colourful tracks, with obstacles for repertoire, and linking the code to Unity
they invent their own 3D game, virtual their marbles to navigate and a goal that projects provides a fun context in which to
world, or visual novel, complete with goals sets off fireworks when it is achieved. make the leap to this new language.
and/or multiple endings.
Example projects are available to play

THESE PROJECTS BUILD A SOLID

“
online, and learners can download the
packages to develop them into their
own creations (for example, helloworld. FOUNDATION OF SKILLS AND CONFIDENCE
cc/crossyrat and helloworld.cc/
marblemayhem). We also encourage
learners to submit their creations for us Learners are encouraged to customise their All the code is provided in the projects,
to showcase to the community on the marble using the open-source Blender along with full explanations, and there are
Projects site. It was fantastic to even see program (a 3D-creation package for loads of examples of how the properties
some creative Unity submissions featured designers and engineers), which enables of a range of game objects can be
in the 2023 Coolest Projects Live showcase Unity developers to enhance the visual manipulated. Learners can then adapt
(helloworld.cc/CP2023), such as a spin on richness and uniqueness of their projects these scripts with ease, providing an almost
the star collecting project with added the- (helloworld.cc/blendermarble). limitless way for them to express their own
floor-is-made-of-lava jeopardy! Progressing through the Explore project creativity and produce something truly
stage, learners create a disco dance floor. individual. The project pathways are both
More Unity This is a truly dynamic learning experience available now at helloworld.cc/introunity
The second pathway, ‘More Unity’, allows that uses tiles to show off colour-changing and helloworld.cc/moreunity. Get exploring,
learners to manipulate in-world physics and materials and trigger audio samples. The designing, and inventing with your young
materials to add depth to their creations. ‘Don’t Fall Through’ puzzle game then creators today and who knows, maybe
We assume some knowledge from the includes learners navigating a safe path they’ll create the next Among Us!

helloworld.cc 9
NEWS FEATURE

n ESA astronaut Andreas Mogensen will help

run Astro Pi code onboard the ISS

THE EUROPEAN ASTRO PI

CHALLENGE 2023/24
© ESA/NASA

The European Astro Pi Challenge is back with exciting missions for 2023/24,
inviting young minds to venture into space through coding

Fergus Kirkpatrick

hat is Astro Pi?

W
and can be delivered in a one-hour session.
The European Astro Pi Challenge
(astro-pi.org) offers young people
It offers a great introduction to text-based
programming with Python. MISSION SPACE LAB
the opportunity to conduct scientific
investigations in space by writing programs Mission Space Lab: 6 November 2023: Registration opens
that run on Raspberry Pi computers on the a new format for 2023/24 19 February 2024: Deadline for submissions
April 2024: Mission ‘flight status’ updates
International Space Station (ISS). We offer Mission Space Lab is changing in 2023/24
June 2024: Certificates sent to teams
two missions for different abilities: Mission to focus more on developing programming
Zero and Mission Space Lab. and problem-solving skills. Working in
teams, young people are invited to come
Mission Zero: suitable for beginners up with ways to calculate how fast the gives your team an incredible chance to
and primary schoolchildren ISS is travelling as it orbits Earth using the capture unique Earth observation photos
Mission Zero allows young people to different sensors and camera on the Astro and data with your team’s program. All
create text-based programs and pixel Pis. Teams need to write a Python program teams whose programs run on the ISS will
art. This year’s theme is ‘flora and fauna’, to capture data and give an output for also receive a personalised certificate and
and successful programs will be sent to their calculation. the data collected in space. All successful
the Astro Pi computers onboard the ISS. To help with this challenge, Mission teams will also be invited to join a Q&A
Mission Zero is accompanied by a project Control has created a new Python library session with an astronaut from the
guide (helloworld.cc/MissionZero2023) that will show data and photos from European Space Agency (ESA).
previous ISS missions. You can download How will your team approach this
and run it on your computer to test your challenge? How could they use the data

MISSION ZERO program without needing an Astro Pi kit.

We’ll also provide a new project guide
from the Astro Pi?
The European Astro Pi Challenge
25 March 2024: Deadline for submissions to walk you through some of the steps 2023/24 offers young coders an out-of-
April 2024: Mission ‘flight status’ updates required for the challenge. this-world experience. Register, code, and
June 2024: Certificates sent to teams If you are successful, your team’s code who knows, maybe your team’s program
may run in space onboard the ISS. This will end up among the stars.

10 helloworld.cc
 NEWS FEATURE

CELEBRATING THE
COMMUNITY: ST JOSEPH’S
SECONDARY SCHOOL, RUSH
We caught up with the team at St Joseph’s, fresh off the success of
another round of Astro Pi, to delve deeper into their passion for exploring
new ways to get learners excited about computing technologies

Sophie Ashford

I
n the heart of rural Rush, Ireland, “I remember the first time I was like, OK, especially once he dropped in that they
St Joseph’s Secondary School has space is cool, was when I watched a movie. might just be the recipients of some very
been on an inspiring journey, thanks to a It was called Interstellar. I [realised] I might exciting photos.
dedicated group of educators and their want to do something like that in my future. “You get to see photos of Earth that
enthusiastic students. So, when I came to [St Joseph’s], I saw nobody has ever seen. This is a unique view
Nestled amidst the picturesque ‘Market coding as a subject and I was like, ‘Mum, just for you … imagine just talking to
Garden of Ireland’, the school caters to a I’ve got to do coding.’” somebody and saying, oh, there’s a picture
diverse student population, ranging from A key driver in encouraging the students of the Amazon. I took that picture when I
children of farmers to newcomers drawn to to give coding a try has been Mr Murray, or was 14. From space,” Danny adds.
the scenic countryside. Danny, as he is fondly referred to by Danny’s mission is to instil in his students
The St Joseph’s leadership team students and staff alike. Danny was the belief that anything is achievable.
recognised the need to adapt and evolve introduced to the importance of engaging Collaborating on Astro Pi projects has
the curriculum in response to the change students with computing technologies enabled teams to uncover their strengths
from an area steeped in agricultural heritage while teaching science at a school in and foster a strong sense of community.
to a region of evolving demographics. England. He attended a Code Club, saw This unity has transcended the
The school’s proactive approach to children building projects with Raspberry classroom, creating a vibrant makers’ culture
teaching coding from First Year (ages 12– Pis, and couldn’t wait to get involved. at St Joseph’s. Today, a dedicated team of
13) has resonated with many students, Growing his knowledge from there, Danny students takes charge of solving tech-
including Kamaya, a member of the switched up his career when he returned related challenges within the school. Deputy
2022/23 Astro Pi cohort, who discovered home to Ireland. He took on the challenge of Principal Darren Byrne says, “Our own
her passion for computer science through helping St Joseph’s expand its computer students actually go class to class repairing
the movie Interstellar. science offering, and leads the charge on all tech issues. So every day, there are four or
IT-related issues. five students going around checking PCs in
After the school introduced mandatory classrooms. They [...] give classes to our
coding taster sessions to all First Year First Year students on app usage.
students, Danny was blown away by the “It’s instilled in the whole school now, the
eagerness of the learners and wanted to idea that the students can look after this
provide further opportunities for them to kind of technology themselves. We’re the
see what they could achieve. ones reaching out for help from
Enter Astro Pi. One of Danny’s the students!”
acquaintances told him about the challenge, You can read more about St Joseph’s
and he shared it with his computer science Secondary School and their story on the
students. In fact, he invited all St Joseph’s Raspberry Pi Foundation website
n A group of St Joseph’s students who took part in the
2022/23 Astro Pi mission students to join in. The uptake was vast, (RPF.io/stjoseph).

helloworld.cc 11
COLUMN
OPINION

ARTIFICIAL INTELLIGENCE
AND THE INCLUSIVE
COMPUTING CLASSROOM
How can you use generative artificial intelligence to
support inclusive practice in computing lessons?

A
rtificial intelligence (AI) is one of the hottest instructions on how to do a certain task, and even specify
topics in education at the moment, and it has a the reading age of the person they are for:
large number of possible benefits for developing
an inclusive classroom. In this article I am going to be Sample prompt: Write me simple step-by-step
looking at the application of generative AI, in particular instructions, understandable by a ten-year-old, on how to
large language models such as ChatGPT and Google’s use a formula to work out the average of a column of data
Bard. As with any new technology, teachers should in Excel.
ensure they are aware of the risks as well as the benefits.
For example, the outputs of generative AI aren’t always Students can then use these to complete an activity, and
accurate; data can be biased, depending on the quality of can even ask the chatbot for help if stuck. Obviously, this
the data it was trained on; and our young people can still present barriers if students struggle with
need to be taught how to recognise the reading, but you can paste the text into
validity of information and to question something like Immersive Reader, or the
whether something may have been NaturalReader web app can be used
generated by AI. Many tools also have a directly in ChatGPT to read the responses
lower user age limit of 13, so make sure aloud (helloworld.cc/NaturalReader).
you check this before letting children use
the tools independently. 2. Support executive function in
Despite the potential drawbacks, there independent tasks
are, however, a number of ways we can Executive function encompasses the skills
use generative AI to support learning, that let us set goals, plan, prioritise tasks,
individualise content, and provide and get things done. Research indicates
creative hooks into topics in computing Catherine Elliott leads the Sheffield that up to 80% of autistic people suffer
eLearning Service (sheffieldclc.net)
to increase engagement. Here are a few from executive function disorder, and
and has spent a number of years
ideas to get you started. other neurodiverse learners may also
working on ways to make computing
struggle in this area. One approach to
accessible for all learners. She is
1. Create step-by-step instructions co-chair of the CAS Include working support students with organising their
Consider a lesson in which students group, and leader of the Sheffield work is to use a ‘body double’; simply
are using a piece of software for the and South Yorkshire Secondary and having another person present when
first time. You can ask an AI chatbot Virtual SEND CAS Communities completing tasks can help them focus
to generate detailed step-by-step (@catherinelliott). better, and an AI chatbot can be used

12 helloworld.cc
to replicate this. There is a great thread about this on X
(formerly Twitter) from an app developer with executive
function disorder (helloworld.cc/EFD).
You could set up a prompt like the one below to keep
a learner on track; or older students could use it to break
down homework or coursework tasks into manageable
steps and keep them focused (see Figure 1):

Sample prompt: You are a friendly, supportive teaching

assistant. You are going to help a child open a document
in Microsoft Word called ‘Poem’ that is saved in the
shared workspace. They then need to insert an image
from the internet to go with the poem, and format the n Figure 1 ChatGPT can be used as a tool to support executive function in independent tasks

poem, e.g. make the title bold. You need to break the task
down into very small steps. Provide step 1 to the child In Google’s Bard you can input your prompt using the
and check they have understood and completed it before microphone, which might be easier for some learners. As
moving on to step 2, etc. with all interactions with generative AI, make sure you
check the accuracy of the outputs!
Again if a learner hasn’t understood a step, they can ask
for clarification or more help. 4. Generate creative prompts
For many learners with SEND, starting with a blank page
is incredibly difficult, both in terms of executive function
and self-confidence. Richard Brown, a teacher at Heritage

I HAVE BEEN USING AI AS

“
Park School in Sheffield, UK, for learners with social,
emotional, and mental health needs (SEMH) has been
A USEFUL SCAFFOLD TO using ChatGPT to generate prompts for creative writing to

MOTIVATE AND SUPPORT engage learners in grammar work. He says, “I have been
using AI as a useful scaffold to motivate and support our
OUR SEN STUDENTS SEN students in their learning; I’ve seen a dramatic impact
on their engagement as a result.”
You could use generative AI in a similar way to create
starting points for a computing project, for example,
3. Simplify complex ideas and content writing a script for a video project that can then be refined
This is something you as a teacher can do, to check the and developed by learners:
reading age of your instructions or worksheets; or a
student could ask an AI chatbot to explain in simpler terms Sample prompt: Please write a short script of no more than
a concept they are struggling with. 400 words for a film about a group of eleven-year-olds who
find an alien creature in their classroom store cupboard.
Sample prompt: Rewrite the following text using
language an eight-year-old can understand … As before, this is just a starting point, and learners
would be expected to adapt and improve it.
You can even ask AI to interrogate a student’s knowledge Generative AI isn’t going to solve all of the issues faced
on a topic, and encourage a conversation around examples by our learners with SEND in the computing classroom.
of a concept in practice: However, used effectively — to provide personalised
learning experiences, give feedback and support in
Sample prompt: You are a friendly, supportive teaching completing tasks, and as a prompt for creative projects
assistant. Find out what the student knows about input — it can help us create a more inclusive, equitable, and
and output devices in computing. inspiring computing classroom.

helloworld.cc 13
RESEARCH

#INSIGHTS
THE SEAME FRAMEWORK: A
SIMPLE WAY TO THINK ABOUT AI
EDUCATION RESOURCES
STORY BY Jane Waite

T
here are many different reasons RPFseminarpapers). Listening to our fellow none at all. We rarely found associated
we all need to learn about artificial researchers, it became clear that resources assessment activities.
intelligence (AI) technologies. They can to teach AI were being developed, but what To analyse the content of the resources,
affect our daily lives, and understanding should be included in these materials was we used a simple framework called
these technologies can open up not yet agreed upon. SEAME. This framework is based on work
opportunities for future careers. So there Therefore, we embarked on a research conducted in 2018 by myself, Jane Waite,
is broad agreement that education about study to review the resources available for with Professor Paul Curzon at Queen
AI is essential. teachers to select from. Our review of over Mary University of London while running
At the Raspberry Pi Computing 300 resources found that the vast majority professional development for educators on
Education Research Centre (helloworld. of learning materials were one-off activities, teaching machine learning (ML) (helloworld.
cc/rpiresearchcentre), we have started and many claimed to be appropriate for cc/teachingML).
investigating the teaching and learning learners of any age. There were very few
of AI. In 2021 and 2022, we hosted a sets of lessons or units of work that were The SEAME framework
series of research seminars on AI and tailored to a specific age group. Activities The SEAME framework gives you a simple
data science education (helloworld.cc/ often had vague learning objectives, or way to view learning objectives and
resources related to teaching AI and ML,
based on whether they focus on social
and ethical aspects (SE), applications (A),
models (M), or engines (E, i.e. how AI works).
We hope it will be a useful tool for other
researchers and educators investigating,
developing, or delivering AI lessons.
The four levels of the SEAME framework
do not indicate a hierarchy or sequence.
Instead, they offer a way to talk about the
focus of AI learning activities.

Social and ethical aspects (SE)

The SE level covers activities that relate to
the impact of AI on everyday life and its
implications for society. Learning objectives
n The SEAME framework and their related resources categorised at

14 helloworld.cc
 representative the data is of all potential
application users (M level).

The SEAME framework as a tool for

research on AI education
For researchers, we think the SEAME
framework will be useful to analyse
curriculum material, for example, to see
whether some age groups have more
learning activities available at one level
than another and whether this changes
over time. We may find that primary-
school learners work mostly at the SE and
A levels, and secondary-school learners
n An SE activity from Experience AI (experience-ai.org) where students explore the social and ethical implications
of an AI application that predicts what subjects they might want to study move between the levels with increasing
clarity as they develop their knowledge. It
this level introduce students to issues, such Covering the four levels may also be the case that some students
as privacy or bias concerns, the impact of Some learning activities will focus on a or teachers prefer activities focused on
AI on employment, misinformation, and the single level, but activities can also span more one level rather than another. However,
potential benefits of AI applications. than one level. For example, an activity may we can’t be sure: research is needed to
start with students trying out an existing investigate the teaching and learning of AI
Applications (A) rock, paper, scissors application that uses and ML across all year groups.
The A level refers to activities related to an ML model to recognise hand shapes.
developing applications and systems that This would cover the application (A) level. If What next?
use AI or ML models. At this level, students students then move on to train the model to We are continuing our research on what
learn how to train models themselves, and improve its accuracy by adding more image should be taught about AI to young people.
look at how such models work. Learning data, they work at the model (M) level. Read more about our findings (helloworld.
objectives at the A level include knowing Other activities cover several SEAME cc/AIpapers); our academic paper that
a range of AI applications and starting to levels to address a specific concept. For describes the SEAME framework and our
understand the difference between rule- example, an activity focused on bias might categorisation of resources will also be
based and data-driven approaches to start with an example of the societal added soon. You can also find out about
developing applications. impact of bias (SE level). Students could the work of PhD student Salomey Afua
then discuss the AI applications they Addo (helloworld.cc/Salomey), who
Models (M) use and reflect on how bias affects them has joined the Raspberry Pi Computing
The M level relates to the models underlying personally (A level). The activity could finish Education Research Centre and is
AI and ML applications. Learning objectives with students exploring related data in a investigating approaches to the teaching
include students understanding the simple ML model and thinking about how and learning of AI.
processes used to train and test models.
For example, through resources focused
on the M level, students could learn
about the different learning paradigms
of ML (i.e. supervised, unsupervised, or
reinforcement learning).

Engines (E)
The E level considers the engines that make
AI models work. This is the most hidden and
complex level, and school-aged students
may need to be taught using unplugged
activities and visualisations. Learning
objectives could include understanding
the basic workings of systems, such as
data-driven decision trees and artificial
n An M-level activity from Experience AI where students learn about classification; although the activity is mostly focused on the M
neural networks. level, it also links to the social and ethical (SE) and application (A) levels

helloworld.cc 15
RESEARCH

ENGAGING CHILDREN
WITH AI ETHICS
STORY BY Katharine Childs

© vejaa/stock.adobe.com
n the context of growing interest
in introducing children and young
people to artificial intelligence (AI)
systems, researchers Mhairi Aitken and
Morgan Briggs from The Alan Turing
Institute (turing.ac.uk) argue it’s crucial
to engage children in ethical discussions
about AI. Aitken and Briggs believe that
children have important rights to consider
when thinking about how AI technology is
designed, developed, and used.

Introducing AI ethics n Children already use AI in their daily lives,

FURTHER READING
The primary goal of AI ethics is to evaluate in technology such as mobile phones

the ethical and social implications of artificial A itken, M. & Briggs, M. (2022).
intelligence technologies. AI ethics is guided Child-centred AI Engaging children with AI ethics. In
by principles rather than rigid rules. These ‘Child-centred AI’ emphasises involving AI, data science, and young people.
principles are encapsulated in the Institute’s children in the entire AI life cycle. This Understanding computing education
SUM values, which have been introduced to approach prioritises informed choices, (Vol 3). Proceedings of the Raspberry
support responsible AI development. As AI children’s involvement in AI discussions, Pi Foundation Research Seminars.
ethics evolve, it is crucial to involve voices ethical awareness among developers, and (helloworld.cc/AitkenBriggs).
that have been underrepresented, such as creating an ethical AI environment.
those of young people. Aitken and Briggs UNICEF collaborated with the enthusiastic about involving children in AI
argue that children’s input is invaluable in Government of Finland in 2020 to create discussions, they lacked effective strategies.
helping shape ethical AI practices, particularly a draft policy guide, Policy Guidance on AI Because children’s active participation in
because AI can potentially have unequal for Children (helloworld.cc/UNICEF), which AI decision-making is crucial, Aitken and
impacts on different segments of society. outlines nine requirements for child-centred Briggs call for innovative approaches to
AI: emphasising child well-being, inclusion, engage children in AI ethics discussions.
fairness, data privacy, safety, transparency,
Exploring AI ethics in the classroom
THE SUM VALUES governmental knowledge, child preparation
for future developments in AI, and creating Drawing on Aitken and Briggs’ ideas,
an enabling environment. engaging children in classroom discussions
The Alan Turing Institute has identified four
The Alan Turing Institute tested UNICEF’s about AI can also introduce ethical
values that support, underwrite, and motivate
draft policy guidance and interviewed 14 considerations. Students can explore topics
(SUM) responsible innovation with new
UK public-sector organisations (helloworld. such as:
technologies such as AI. The values are:
cc/Turingcasestudy). The findings revealed n B ias in AI algorithms (helloworld.cc/

n R
 espect the dignity of individuals challenges in implementing child-centred MLbias)
n C
 onnect with each other sincerely, openly,
AI, such as low data literacy among the n S afety concerns (helloworld.cc/AIsafety)

and inclusively public, a lack of understanding of GDPR n R esponsible use of AI (helloworld.cc/

issues, and concerns about AI guidance. ResponsibleAI)

n C
 are for the well-being of each and all
Organisations expressed a desire for By addressing these issues, teachers help
n P
 rotect the priorities of social values,
actionable recommendations, sector- students become responsible and ethical
justice, and the public interest
specific guidance, and content accessible users of AI, contributing to a more equitable
to all ages. While organisations were and just society.

16 helloworld.cc
 PRODABI: THE PROJECT ON DATA
SCIENCE AND BIG DATA
STORY BY Claire Johnson

A
s the march of AI gathers pace,
educationalists are carefully
(helloworld.cc/Rahwan). The paper called
for a new discipline, ‘artificial science’, to
FURTHER READING
considering what knowledge secondary study intelligent machines as “a class of
R
 ahwan, I., et al. (2019). Machine
students (aged 11–18) need to develop actors with particular behavioural patterns
Behaviour. Nature, 568(7753), 477–
and how teachers can effectively deliver it. and ecology, not just engineering artefacts”.
486. helloworld.cc/Rahwan
To this end, since 2018, academics at This is the scientific study of machine and
Paderborn University in Germany have human–machine ecologies. S chulte, C., and Budde, L.: A
been developing ProDaBi, a research-based These ‘hybrid interaction systems’ Framework for Computing Education:
curriculum for secondary-school level in (helloworld.cc/Schulte2018) in turn call for Hybrid Interaction System: the need
data science and big data (including AI). a focus on the interaction between human for a bigger picture in computing
Carsten Schulte and his team suggest and machine in an effective CS curriculum, education. In Proceedings of the 18th
that AI and machine learning (ML) differ but also for a greater focus on the central Koli Calling International Conference
from other topics in the computer science role of data. This approach forms the basis on Computing Education Research
(CS) curriculum and might benefit from a of the teaching material developed over the (2018). helloworld.cc/Schulte2018
data-centric approach to complement the two phases of the ProDaBi project.
traditional focus on programming: “These
topics require a paradigm shift for some Teaching materials vs Machine!’, introduces students to ML
teachers, and ... this shift has to do with the The research initially generated an optional, by illustrating how a simulated artificial
changed role of algorithms, data, and the year-long course for 17- to 18-year-olds, in agent gets better at playing the fictional
societal context.” which they learn about data exploration and game Hexapawn (helloworld.cc/
Their position is partly informed ML. Successive evaluations led to teaching humanvmachine).
by a 2019 paper, Machine Behaviour materials for 15- to 16-year-olds, which build
data awareness over four units as follows: Role play
1. ‘Data science and artificial intelligence The ProDaBi project urges CS teachers to

HYBRID INTERACTION SYSTEM (without programming)’: students

analyse survey data on young people’s
pay more attention to AI technologies and
focus on the interaction between humans
media use and learn about decision and digital artefacts and the transformational
“AI education requires developing an adequate
trees as data-based models, using effects of these hybrid systems. As we are
picture of the hybrid interaction system — a
Common Online Data Analysis Platform both shaped by and shape these data-driven
kind of data-driven, emergent ecosystem which
(CODAP) software. technologies, an effective CS curriculum
needs to be made explicit to understand the
2. ‘Data science and artificial intelligence should firmly anchor these considerations to
transformative role as well as the technological
basics of these artificial intelligence tools and
(with programming)’: students learn its AI and ML offerings. The ProDaBi project
how they are related to data science” (Schulte how decision trees and artificial neural teaching materials provide valuable support
et al., 2021. helloworld.cc/hybrid). networks are used in image recognition, here, although not all are available in English.
using Jupyter Notebook with Python.
3. ‘Data awareness’: students consider Research to practice
the role of data used by mobile Teachers in different cultural and
phone networks and movie curriculum contexts may need to adapt
recommendation systems. the extended activities to suit their own
4. ‘Data projects and data exploration’: timescales, or use the unit overviews as a
students collect and analyse source of inspiration for their own lesson
environmental data with sensors, using planning. However you use the materials,
n Hybrid interaction system, adapted from Schulte Arduino and openSenseMap. we would love to hear about it. Email us at
An additional, unplugged activity, ‘Human [email protected].

helloworld.cc 17
INTERVIEW
OPINION

MACHINE LEARNING FOR KIDS

“I HOPE THAT PEOPLE DON’T LOOK AT AI JUST AS
SOMETHING THAT MIGHT REPLACE THEM …”
© CC-BY 4.0

working on IBM Watson products, IBM’s suite of AI tools,

Ben Garside speaks with Dale Lane,
and he was involved in projects that explored whether
chief architect at IBM and founder of computers could be creative. “I’ve done the weird and
wacky stuff, and I’ve done the very cool, pretty stuff,” Dale
Machine Learning for Kids, to discuss
summarises. One project he worked on was called ‘Chef
developments in AI and what they could Watson’, with the goal of building a tool that could learn
from recipes. He describes the idea thus: “If we take [data
mean for education
about what food you buy], can we invent a new meal that
you might like with that combination of food?”
In the not so distant past, a distinction would have been

“I
still think of myself as a software developer, but drawn between a regular software developer and someone
in practice I don’t get to write as much code as I who works in AI, but according to Dale, that distinction
used to,” Dale explains when talking about his has now disappeared. “Nowadays, I think there’s virtually
current job. Having worked at IBM for the last 20 years, he no software product IBM makes that doesn’t have AI
has seen AI develop from an emerging technology to a [technology] in it somewhere ... Every software team has
normal part of all software projects at IBM. “I’ve always a plan of how they’re applying AI [systems] to improve
liked working in areas of tech that are new and cutting- their product; every developer, no matter what part of the
edge,” he says. Back in the early 2010s, he started business they work in, has to understand how to apply AI.”

18 helloworld.cc
 IF I CAN GET A
“ NON-AI PERSON TO
UNDERSTAND AI TECH,
COULD I DO THAT WITH
A NINE-YEAR-OLD?

and allows users to build their own applications based on

machine learning models they have trained themselves.
Dale’s hope is that young people play around with the tool
and as a by-product, “learn a bit about how these projects
are made and what they can be used for”.
Dale tells me that he sees a lot more traffic to the
Machine Learning for Kids website now than when he
launched the tool in 2017. This may not be a surprise —
AI seems to be in many more people’s consciousnesses
than it was even at this time last year. Stories in the news
and the release of AI chatbots based on large language
models (LLMs), such as ChatGPT and Bard, have caught
the attention of the world.
I ask Dale what impact he thinks LLMs will have on the
educational landscape. “I hope a positive one. I’m always
an optimist when it comes to tech. I’m always on the
side of the fence that it’s going to be fine. I do see a huge
amount of value in [LLMs]. I get that there is a lot of fear
around plagiarism, and that’s not unique to education.”
Dale suggests that AI chatbots can be considered like a
“colleague that you ask five questions of, and twice they’ll
n Dale Lane speaking at TEDx Winchester, UK
tell you stuff you already knew, twice they’ll give you a
new idea that gets you thinking, and once they’ll make
AI tools in education something up that’s just complete fantasy. But they’ll do
As new AI systems developed, Dale saw the need to help it with such confidence that you might not notice. It’s still
educate other colleagues about how they worked. “We useful to ask that person, but [you need to] be able to
were trying to enable [non-AI] developers to use AI tech, recognise the one in five times when it’s fantasy.”
so I built a whole load of stuff that tried to build machine This emphasises the importance of educators continuing
learning into a tool so that any developer could use it.” He to focus on something we’re already teaching across
fondly describes how he took the same approach when various subjects: the skills young people need in order to be
he was trying to explain to his children what he did for a able to look critically at secondary sources.
living. “If I can get a non-AI person to understand AI tech, Dale describes a school social media post he’s seen
could I do that with a nine-year-old?” So in his spare time, in which students were challenged to find the mistakes
Dale began working on a tool that would help his children in an essay generated by a LLM. “[The teachers] were
understand, as his explanations had never done before. partly trying to show the kids how it worked and what it
From this, the Machine Learning for Kids website was born. can do, but made sure that there were some real glaring
“Machine Learning for Kids is trying to get people errors in there. Trying to get the kids to recognise both
thinking about how AI [tech] is all around them, the impact the opportunities and the limitations of the tech. I thought
it has on their lives, what the tech can do, and how it’s this was awesome.” Dale feels people shouldn’t buy into
used and behaves ... through hands-on experimenting.” “this idea that we can just describe [AI tools] as bad, only
The Machine Learning for Kids tool integrates with Scratch for cheaters and lazy kids, and that there’s some virtue in

helloworld.cc 19
INTERVIEW
OPINION

problem for society, and Dale believes that deepfakes —

WHAT HAS ENABLED videos generated or manipulated using AI tools — pose

THE RECENT RAPID a unique challenge. “We trust video in a way that we
don’t trust pictures. Before the days of Photoshop, we

DEVELOPMENTS IN AI? would have trusted a photo, but we know photos can be
manipulated now and we look critically at photos before
“The biggest thing [over the past 20 years] is that hardware we trust them. We still trust video, but there are loads of
has come on so much. We used to have to build racks of examples around of videos of famous people looking like
servers. Nowadays, you can [work on AI models] on your they’re saying something that they would never dream
laptop. The advent of cloud means if I do need to train of saying. I could totally see [AI technology] being used
a model that’s a bit complicated, I can spin up a bunch in a way that shows some world leader making some
of servers for a few hours to do it in without needing to big announcement and the world panics before anyone
physically build a rack of servers. All of our computers have realises it was a fake.”
become so much more powerful, sometimes we don’t even The question that governments, civil society,
need servers.” researchers, and industry leaders are asking is how we
control the direction in which AI technology is developed.
I ask Dale what role he sees legislation playing in this,
doing everything the hard or manual way”. He recalls an and he points to the problem
analogy he likes to use when describing advances in AI: “I posed by the wide availability
remember when I was at school, my dad gave me grief for of AI tools. “We do need
using a calculator. In his day, they did mental arithmetic. oversight and legislation,
He would say how my mental arithmetic wasn’t as good but if you’ve got big enough
as his generation’s, which is true, it sucks! But the sums computer or server access,
I was doing were way more complex than what he was you can run these models
doing when he was at school. The calculator was doing the yourself. The days that you
mechanical rope work, freeing me up to do the bigger, more n Machine Learning for Kids gives a could just target a few big
hands-on intro to machine learning
challenging things.” companies with legislation
Part of our conversations in and control it are gone. This
the classroom could relate to stuff is available. These
AI IS NOT GOING TO
“
analogies around AI such as models are open-sourced,
Dale’s. Asking your learners and [people who are planning
to discuss how using an JUST DISAPPEAR to use AI tools] maliciously
AI chatbot to generate an are never going to follow the
essay is like or unlike using legislation. The best you can
a calculator to solve a maths problem could spark them to do is put safeguards in place in the rest of society.”
think more carefully about how they use AI tools. As Dale describes, AI is not going to just disappear from
our lives. But our job as educators remains the same: to
The future with AI help young people navigate their lives as best they can —
What does the future of AI hold for us? Ever optimistic now and in the future. So we need to put effort and time
about technology, Dale says, “I hope that people don’t look into preparing them to be discerning, critical thinkers who
at AI just as something that might replace them, but as understand how to use and assess new technologies as
something that will enable them to do bigger, better, more they emerge.
amazing things, and that’s been true about tech for years.” Machine Learning for Kids is an excellent tool for getting
While there are many potential positives around recent young people started with understanding and building their
AI developments, to be in a place where we’re able to own machine learning models and AI applications. It is the
harness it to people’s benefit, we also must recognise main tool used in our Experience AI Lessons, and the
that it can be used in dangerous ways. When I ask Dale Machine Learning for Kids website also comes with lots of
to share his perspective, he tells me, “I think the biggest worksheets and example projects to get you started.
risks … that we see today come from AI being applied in
ways that people didn’t realise was possible. Not just in For more information about Machine Learning for Kids,
terms of generating text, but if you think of deepfakes.” visit machinelearningforkids.co.uk
Not knowing what media and information to trust is a

20 helloworld.cc
LISTEN TO OUR
PODCAST
Dive into the Hello World
Subscribe to the podcast!
podcast for:
• Even more of the great Hello World • Apple Podcasts
helloworld.cc/applepod
content you love between issues
• Spotify
• First-hand knowledge directly from the helloworld.cc/spotifypod
educators behind our articles • Google Podcasts
helloworld.cc/googlepod
• Learning and expertise as you delve a • Amazon Music
little deeper and listen on the move helloworld.cc/amazonpod
SCAN ME

OUR MOS
POPULAR T
How mora
EPISODES
l is your m
computin achin e? E
g educati thics in
Accessib on
le and in
educatio clu s ive comp
n: where uting
to start?
How can
we get ev
eryone e
about co xcited
de?
SUBSCRIBE FREE
IN P RINT
TODAY for UK-
ba
educat sed
ors

Why subscribe?
• Teaching resources and ideas used by
over 90 percent of our readers
• Exclusive news, research findings,
and in-depth features
• Delivered to your door three times a year

TO SUBSCRIBE VISIT:
helloworld.cc/subscribe

Not a UK-based educator?

FREE
• Buy a print copy from
helloworld.cc/buy — we ship to
over 50 countries
• Subscribe to receive the free
PDF on the day it is released
PDF one,
for any re
e
• Read features and news at anywh
helloworld.cc
© Parradee/stock.adobe.com

n Recent AI developments are causing

a rethink in computing education

WHAT DOES AI MEAN FOR

COMPUTING EDUCATION?
Philip Colligan engages with the big questions
surrounding AI and computing education

I
t’s been less than a year since At the same time, there are those saying 3. Enhancing teaching and learning
ChatGPT catapulted generative it’s all moving too fast; that regulation isn’t through AI-powered technologies
artificial intelligence (AI) into the keeping pace with innovation. One of the
mainstream public consciousness, UK’s leading AI entrepreneurs, Mustafa AI literacy: what it is and how we
reigniting the debate about the role that Suleyman, said recently: “If you don’t start teach it
these powerful new technologies will from a position of fear, you probably aren’t For young people to thrive in a world that is
play in all of our futures. paying attention.” being transformed by AI systems, they need
‘Will AI save or destroy humanity?’ might What does all this mean for education, to understand these technologies and the
seem like an extreme title for a podcast, and particularly for computing education? role they could play in their lives.
particularly if you’ve played with these Is there any point trying to teach children The first problem is defining what AI
products and enjoyed some of their obvious about AI when it is all changing so fast? literacy actually means. What are the
limitations. The reality is that we are still Does anyone need to learn to code anymore? concepts, knowledge, and skills that it would
at the foothills of what AI technology can Will teachers be replaced by chatbots? Is be useful for a young person to learn?
achieve (think World Wide Web in the assessment as we know it broken? In the past couple of years there has been
1990s), and lots of credible people are If we’re going to seriously engage with a huge explosion in resources that claim to
predicting an astonishing pace of progress these questions, we need to understand that help young people develop AI literacy. Our
over the next few years, promising the we’re talking about three different things: research team mapped and categorised
radical transformation of almost every 1. AI literacy: what it is and how we teach it over 500 resources and undertook a
aspect of our lives. Comparisons with the 2. Rethinking computer science (and systematic literature review (helloworld.
Industrial Revolution abound. possibly some other subjects) cc/AIreview) to understand what research

helloworld.cc 23
 FEATURE

has been done on K–12 AI classroom that doesn’t solve the problem of how Rethinking computer science
interventions (spoiler: not much). we get them into the hands and minds Alongside introducing AI literacy, we also
The reality is that — with a few notable of every young person. This will require need to take a hard look at computer science.
exceptions — the vast majority of AI literacy policy change. We need governments At the very least, we need to make sure that
resources available today are probably and education system leaders to grasp computer science curricula include machine
doing more harm than good. For example, that a foundational understanding of learning models, explaining how they
in an attempt to be accessible and fun, AI technologies is essential for creating constitute a new paradigm for computing,
many materials anthropomorphise AI economic opportunity, ensuring that young and give more emphasis to the role that data
systems, using human terms to describe people have the mindsets to engage will play in the future of computing. Adding
them and their functions, perpetuating positively with technological change, and anything new to an already packed computer
misconceptions about what AI systems are avoiding a widening of the digital divide. science curriculum means tough choices
and how they work. We’ve messed this up before with digital about what to deprioritise to make space.
What emerged from this work at the skills. Let’s not do it again. And, while we’re reviewing curricula,
Raspberry Pi Foundation is the SEAME More than anything, we need to what about biology, geography, or any of
model, which articulates the concepts, invest in teachers and their professional the other subjects that are just as likely to be
knowledge, and skills that are essential
ingredients of any AI literacy curriculum (see
THE REALITY IS THAT WE ARE STILL AT THE
“
pages 14–15 in this issue). It separates out
the social and ethical, application, model, and
engine levels of AI systems — all of which FOOTHILLS OF WHAT AI CAN ACHIEVE
are important — and gets specific about
age-appropriate learning outcomes for each.
This research has formed the basis of development. While there are some fantastic revolutionised by big data and AI? As part of
Experience AI (experience-ai.org), a suite computing teachers with computer science Experience AI, we are launching some of the
of resources, lessons plans, videos, and qualifications, the reality is that most of the first lessons focusing on ecosystems and AI,
interactive learning experiences created by computing lessons taught anywhere on the which we think should be at the heart of any
the Raspberry Pi Foundation in partnership planet are taught by a non-specialist teacher. modern biology curriculum.
with Google DeepMind, which is already That is even more so the case for anything There is already a lively debate about
being used in thousands of classrooms. related to AI. If we’re serious about AI literacy the extent to which the new generation of
Defining AI literacy and developing for young people, we have to get serious AI technologies will make programming
resources is part of the challenge, but about AI literacy for teachers. as we know it obsolete. In January, the
prestigious ACM journal ran an opinion
piece from Matt Welsh, founder of an AI-
powered programming start-up, in which
he said: “I believe the conventional idea of
‘writing a program’ is headed for extinction,
and indeed, for all but very specialised
applications, most software, as we know
it, will be replaced by AI systems that are
trained rather than programmed.”
With GitHub (now part of Microsoft)
claiming that their pair programming
© Monkey Business/stock.adobe.com

technology, Copilot, is now writing 46

percent of developers’ code, it’s perhaps
not surprising that some are saying young
people don’t need to learn how to code.
It’s an easy political soundbite, but it just
doesn’t stand up to serious scrutiny.
n Teachers need support and
Even if AI systems can improve to the
professional development
point where they generate consistently

24 helloworld.cc
reliable code, it seems to me that it is just are, none of them have even begun to
as likely that this will increase the demand harness the potential of AI systems like
for more complex software, leading to large language models (LLMs).
greater demand for more programmers. One area where I think we’ll see huge
There is historical precedent for this: progress is feedback. It’s well established
the invention of abstract programming that good-quality feedback makes a huge
languages such as Python dramatically difference to learning, but a teacher’s ability
simplified the act of humans providing to provide feedback is limited by their time.
instructions to computers, leading to more No one is seriously claiming that chatbots
complex software and a much greater will replace teachers, but — if we can get
demand for developers. the quality right — LLM applications could
However these AI-powered tools provide every child with unlimited, on-
develop, it will still be essential for demand feedback. AI powered feedback
young people to learn the fundamentals — not giving students the answers, but
of programming and to get hands-on coaching, suggesting, and encouraging in
experience of writing code as part of any the way that great teachers already do — PHILIP COLLIGAN
credible computer science course. Practical could be transformational. Philip is the chief executive of the Raspberry
experience of writing computer programs We are already seeing edtech companies Pi Foundation, an educational nonprofit with
is an essential part of learning how to racing to bring new products and features a mission to enable young people to realise
analyse problems in computational terms; to market that leverage LLMs, and my their full potential through the power of
it brings the subject to life; it will help prediction is that the pace of that innovation computing and digital technologies.
young people understand how the world is going to increase exponentially over the
around them is being transformed by AI coming years. The challenge for all of us
systems; and it will ensure that they are working in education is how we ensure as an AI kitemark for edtech products that
able to shape that future, rather than it that ethics and privacy are at the centre communicate whether they meet a set of
being something that is done to them. of the development of these technologies. standards around bias, transparency,
and privacy.

HOW MUCH DATA FROM STUDENTS WILL Realising potential in a brave

“ AN AI SYSTEM NEED TO ACCESS?
new world
We may very well be entering an era in
which AI systems dramatically enhance the
creativity and productivity of humanity as a
Enhancing teaching and learning That’s important for all applications of species. Whether the reality lives up to the
through AI-powered technologies AI, but especially so in education, where hype or not, AI systems are undoubtedly
Technology has already transformed these systems will be unleashed directly going to be a big part of all of our futures,
learning. YouTube is probably the most on young people. How much data from and we urgently need to figure out what
important educational innovation of the students will an AI system need to access? that means for education, and what skills,
past 20 years, democratising both the Can that data — aggregated from millions knowledge, and mindsets young people
creation and consumption of learning of students — be used to train new models? need to develop in order to realise their full
resources. Khan Academy, meanwhile, How can we communicate transparently potential in that brave new world.
integrated video instruction into a learning the limitations of the information provided That’s the work we’re engaged in at the
experience that gamified formative back to students? Raspberry Pi Foundation, working in
assessment. Our own edtech platform, Ultimately, we need to think about how partnership with individuals and
Ada Computer Science, combines parents, teachers, and education systems organisations from across industry,
comprehensive instructional materials, a (the purchasers of edtech products) will be government, education, and civil society.
huge bank of questions designed to help able to make informed choices about what If you have ideas and want to get involved
learning, and automated marking and to put in front of students. Standards will in shaping the future of computing
feedback to make computer science easier have an important role to play here, and I education, we’d love to hear from you at
to teach and learn. Brilliant though these think we should be exploring ideas such [email protected].

helloworld.cc 25
 FEATURE

HOW ANTHROPOMORPHISM
HINDERS AI EDUCATION
Ben Garside explains why talking about artificial intelligence
systems as sentient beings is a bad idea

I
n the 1950s, Alan Turing explored The reality is that these new chatbots
the central question of artificial are applications based on a large language
intelligence (AI). He thought that finding model (LLM) — a machine learning model
the answer to a question which many trained with vast quantities of text, written
researchers were discussing at the by people, and taken from places such as
time, ‘Can machines think?’, would not books and the internet. An LLM predicts the
provide valuable answers because the probable order of combinations of words,
terms ‘machine’ and ‘think’ are hard to a bit like the autocomplete function on a
define. Instead, he proposed changing the smartphone. Based on these probabilities, it
question to something more provable: ‘Can can produce text outputs. LLM chatbots run
a computer imitate intelligent behaviour on servers with huge amounts of computing
well enough to convince someone they power that people have built in data centres
are talking to a human?’ (helloworld.cc/ around the world.
BEN GARSIDE
Ben is a learning manager at the Raspberry Pi
turing1950). This is commonly referred to
Foundation. He has worked on the production
as the Turing test. The problem with anthropomorphism
of The Computing Curriculum and has written
It’s been hard to miss the newest AI applications are often described as
online courses for educators, including
generation of AI chatbots that companies ‘black boxes’ or ‘closed boxes’; they may be
Introduction to Machine Learning and AI.
have released over the last year. News relatively easy to use, but it’s not as easy to
articles and stories about them seem to understand how they work. As educators, it’s
be everywhere, so you may have heard of fundamentally important to help everyone, or treating an object such as an AI system
machine learning (ML) chatbots such as especially young people, understand the as though it were human in appearance,
ChatGPT and Bard. These chatbots are potential of AI technologies and open these character, or behaviour) is a term that
advanced enough to have caused renewed closed boxes to know how they actually comes up regularly in conversations with
discussions about the Turing test and work. We should be demystifying digital my colleagues when we are developing
whether the chatbots are sentient. technology for young people to empower resources at the Raspberry Pi Foundation.
them to be thoughtful creators of technology Anthropomorphising AI in teaching
Chatbots are not sentient and to make informed choices about how materials might lead learners to believe
Without any knowledge of how people they engage with technology — rather than that there is sentience or intention within
create such chatbots, it’s easy to imagine just being passive consumers. AI applications. That misconception would
how someone might develop an incorrect To support learners in forming accurate distract learners from the fact that it is
mental model around these chatbots being mental models of AI and ML systems, it is people who design AI applications and
living entities. With some awareness of important to avoid using words that can lead decide how they are used. It also risks
sci-fi stories, you might even start to to learners developing misconceptions about reducing learners’ desire to take an active
imagine what they could look like or machines being humanlike in their abilities. role in understanding AI applications and
associate a gender with them. That’s why anthropomorphism (showing designing future applications.

26 helloworld.cc
 © lensw0rld/stock.adobe.com
in the area: artificial intelligence, machine

IT’S MORE ACCURATE TO SAY THAT AI learning, and computer vision, to name

“ RECEIVES INPUT AND PROCESSES THE DATA

but a few examples. At the Raspberry Pi
Foundation, we are still training ourselves
not to anthropomorphise AI, and we take a
bit of pleasure in picking each other up on
Avoiding anthropomorphism helps differently from how a human might identify the odd slip-up.
young people to open the closed box of animals in images. Where humans would If we are correct in our approach when
AI applications. Take the example of a point to features such as whiskers and ear developing resources at the Raspberry Pi
smart speaker. It’s easy to describe a smart shapes, ML models process pixels in images Foundation, then whether or not the young
speaker’s functionality in anthropomorphic to make predictions based on probabilities. people who use our resources grow up to
terms such as ‘it listens’ or ‘it understands’. become AI developers, we will have helped
However, it’s more accurate and empowering Avoiding the AI ‘personality’ trap them to become discerning users of AI
to explain smart speakers as systems It’s not easy to avoid anthropomorphism technologies. They will be more likely to see
developed by people to process sound and while talking about AI, especially such products for what they are: data-driven
carry out specific tasks. Rather than telling considering the industry-standard language applications and not sentient machines.
young people that a smart speaker ‘listens’
and ‘understands’, it’s more accurate to say
that the speaker receives input, processes the
data, and produces an output. This language
BETTER WAYS TO DESCRIBE AI
helps to distinguish how the device actually
works from the illusion of a persona the Terms to avoid Terms to use
speaker’s voice might conjure for learners. Avoid using phrases such as ‘AI learns’ or Use phrases like ‘AI applications are designed to’ or
Another example is the use of AI in ‘AI/ML does’ ‘AI developers build applications that’
computer vision (a field of AI that attempts
to gain meaningful information from images Don’t use words that describe the behaviour of Prefer system-related words (e.g. detect, input,
and videos). ML models can, for example, be people (e.g. see, look, recognise, create, make) pattern-match, generate, produce)
trained to identify when there is a dog or a
cat in an image. An accurate ML model, on Avoid using AI/ML as a countable noun, e.g. ‘new Refer to ‘AI/ML’ as a scientific discipline, similar to
the surface, therefore displays humanlike artificial intelligences emerged in 2022’ how you use the term ‘biology’
behaviour. However, the model operates very

helloworld.cc 27
 FEATURE

n Existing AI facial recognition

systems do not identify
dark-skinned and female
faces accurately

© Leszek/stock.adobe.com
BIAS IN THE MACHINE: HOW CAN
WE ADDRESS GENDER BIAS IN AI?
Sue Sentance examines the intersection of AI and gender, particularly
with respect to gender bias in machine learning

G
ender inequality is everywhere, and even crash test dummies are all based In order for a ML model to categorise
and researchers, activists, and on data gathered from men. new data appropriately, it needs to be
governments themselves, have struggled Looking at this problem from a different trained with data that is gathered from
since the 1960s to tackle it. As women and angle, researcher Mayra Buvinic and her everyone, and is, in the case of supervised
girls around the world continue to suffer colleagues highlight that in most countries learning, annotated without bias. Failing
from discrimination, the United Nations of the world, there are no sources of data to do this creates a biased ML model. Bias
has pledged to achieve gender equality that capture the differences between has been demonstrated in different types
and to empower all women and girls male and female participation in civil of AI systems that have been released as
through its Sustainable Development Goals society organisations, or in local advisory products. For example:
(helloworld.cc/UN). or decision-making bodies. A lack of data n Facial recognition: AI researcher Joy
While progress has been made, new about girls and women will surely have a Buolamwini discovered that existing AI
developments in technology may be negative impact on decision-making. facial recognition systems do not identify
threatening to undo this. As Susan Leavy, dark-skinned or female faces accurately.
a machine learning (ML) researcher from Bias in machine learning Her discovery, and her work to push
the Insight Centre for Data Analytics, Machine learning is a type of AI technology for the first-ever piece of legislation
puts it in her conference proceedings, that relies on vast data sets for training. ML in the USA to govern against bias in
“Artificial intelligence is increasingly is currently being used in various systems for the algorithms that impact our lives,
influencing the opinions and behaviour automated decision-making. Bias in data sets is narrated in the 2020 documentary
of people in everyday life. However, the (helloworld.cc/biasindata) for training ML Coded Bias (codedbias.com).
overrepresentation of men in the design models can be caused in several ways. For n Natural language processing: imagine
of these technologies could quietly undo example, data sets can be biased because an AI system that is tasked with filling
decades of advances in gender equality.” they are incomplete or skewed (as is the case in the missing word in ‘Man is to king
in data sets which lack data about women). as woman is to X’ and comes up with
Gender-biased data Another example is that data sets can be ‘queen’. But what if the system completes
In her 2019 award-winning book Invisible biased because of the use of incorrect labels ‘Man is to software developer as woman
Women: Exposing Data Bias in a World by people who annotate the data. Annotating is to X’ with ‘secretary’ or some other
Designed for Men, Caroline Criado Perez data is necessary for supervised learning, word that reflects stereotypical views
discusses the effects of gender-biased where ML models are trained to categorise of gender and careers? AI models called
data. She describes, for example, how the data into categories decided upon by people ‘word embeddings’ learn by identifying
designs of cities, workplaces, smartphones, (e.g. pineapples and mangoes). patterns in huge collections of texts. In

28 helloworld.cc
 speakers with longer vocal cords and lower- reduce gender bias. A 2020 report by the
FURTHER READING pitched voices (typically males). World Economic Forum (WEF) on gender
B
 uvinic, M., & Levine, R. (2016). Closing The article recommends four ways for the parity found that women account for only
the gender data gap. Significance, ML workforce to try to avoid gender bias: 26 percent of data and AI positions in the
13(2):34–37. helloworld.cc/Buvinic n Ensure diversity in the training data (using workforce (helloworld.cc/gendergap).
the example from the article, including The WEF suggests five ways in which
L eavy, S. (2018). Gender bias in artificial
audio samples from all genders, not just the AI workforce gender balance could
intelligence: The need for diversity and
male) be addressed (helloworld.cc/WEF):
gender theory in machine learning. In
n Ensure that a diverse group of people 1. Support STEM education
Proceedings of the 1st International
Workshop on Gender Equality in Software labels the training data 2. Showcase female AI trailblazers
Engineering (GE 2018) (pp. 14–16). n Measure the accuracy of a ML model 3. Mentor women for leadership roles
helloworld.cc/Leavy separately for different demographic 4. Create equal opportunities
categories to check whether the model is 5. Ensure a gender-equal reward
P erez, C. C. (2019). Invisible Women: biased against some of the categories system
Exploring Data Bias in a World Designed n Establish techniques to encourage ML n Ensure the collection of and access
for Men. Random House. helloworld. models towards unbiased results to high-quality and up-to-date
cc/Perez
gender data. We need high-quality
What can everybody else do? data sets on women and girls, with
addition to the structural patterns of the The above points can help people in the AI good coverage, including country
text language, word embeddings learn industry, which is of course important — but coverage. Data needs to be comparable
human biases expressed in the texts. You what about the rest of us? It’s important to across countries in terms of concepts,
can read more about this in the Brookings raise awareness of the issues around gender definitions, and measures. Data
Institute report (helloworld.cc/Brookings). data bias and AI lest we find out too late should have both complexity and
that we are reintroducing gender inequalities granularity, so it can be cross-tabulated
Not noticing we have fought so hard to remove. and disaggregated, following the
There is much debate about the level of bias Awareness is a good start, and some other recommendations from the Data2x
in systems using AI, and some researchers suggestions, drawn out from others’ work in project on mapping gender data gaps
worry that this will cause distrust in ML this area, are: (helloworld.cc/data2x).
systems. Thus, some scientists are keen to n Improve the gender balance in the n Educate young people about AI.
emphasise the breadth of their training data AI workforce. Having more women in At the Raspberry Pi Computing Education
across genders (helloworld.cc/ultromics). AI and data science, particularly in both Research Centre, we believe that
However, other researchers point out technical and leadership roles, will help to introducing some of the potential (positive
that despite all good intentions, gender and negative) impacts of AI systems
disparities are so entrenched in society to young people in school may help to
that we are not aware of all of them. White build awareness and understanding at a
and male dominance in our society may be young age. In our AI education seminars
so unconsciously prevalent that we don’t we heard a number of perspectives
notice all its effects. on this topic, and you can revisit the
videos, presentation slides, and blog
What is being done in the AI sector to posts (helloworld.cc/AIseries). We’ve
address bias? also conducted a literature review of
Developers and researchers of AI systems AI education interventions in K–12
have been trying to establish rules for how (helloworld.cc/AIreview).
to avoid bias in AI models. An example rule SUE SENTANCE
set is given in a Harvard Business Review Sue is director of the Raspberry Pi We are using our understanding of and
article (helloworld.cc/Harvard), which Computing Education Research Centre. research around AI to develop new learning
describes the fact that speech recognition An ex-teacher, a teacher-trainer, and programs. As we think deeply about how AI
systems originally performed poorly for an academic, she is passionate about concepts can be learnt and understood, we
non-male speakers, because systems research in computing education. will continue to educate young people about
analysed and modelled speech for taller the risks of bias in AI.

helloworld.cc 29
 FEATURE

EXPERIENCE AI: MAKING AI

RELEVANT AND ACCESSIBLE
Google DeepMind’s Aimee Welch discusses the Experience AI programme and
why equal access to AI education is key

F
rom chatbots to self-driving cars, make more informed decisions about using diverse values, hopes, and concerns into the
AI is here to stay, and it’s rapidly AI applications in their daily lives, helping design and deployment of the technology.
transforming our world. It can potentially ensure safe and responsible use of the This is critical for AI to benefit everyone
solve some of the biggest challenges technology. The UK government’s AI Council (helloworld.cc/equitableAI).
humanity faces today. But it also has many acknowledges this and aims to have every By focusing on AI education from a young
serious risks and inherent challenges, like child in the UK leave school with a basic age, there is an opportunity to break down
reinforcing existing patterns of bias or understanding of how AI works, as stated in some of these long-standing barriers. That’s
‘hallucinating’, a term that describes AI its AI Roadmap (helloworld.cc/AIroadmap). why we partnered with the Raspberry Pi
making up false outputs that do not reflect But while AI literacy is a key skill in Foundation to co-create Experience AI,
real events or data. this new era, not every young person a new educational programme with free
As AI becomes integral to our daily currently has access to sufficient AI lesson plans, slide decks, worksheets, and
lives, younger generations must gain the education and resources. In a recent survey videos, to address gaps in AI education and
knowledge and skills to navigate and by the EdWeek Research Center in the support teachers in engaging and inspiring
shape this technology. Young people with US (helloworld.cc/EdWeekAIsurvey), young people in the subject.
a foundational understanding of AI can only one in ten teachers said they knew The programme aims to help young
enough about AI to teach its basics. Very people aged 11–14 take their first steps
few reported receiving any professional in understanding the technology, making
development related to the topic. it relevant to diverse learners, and
Similarly, our work with the Raspberry Pi encouraging future careers in the field. All
Computing Education Research Centre Experience AI resources are freely available
(computingeducationresearch.org) has to every school across the UK and beyond.
suggested that UK-based teachers are The partnership is built on a shared
eager to understand more about AI and vision to make AI education more inclusive
how to engage their students in the topic. and accessible. Bringing together the
Foundation’s expertise in computing
Bringing AI into classrooms education and our cutting-edge technical
Ensuring broad access to AI education is knowledge and industry insights has
also important. This way, we can improve allowed us to create a holistic learning
diversity among people working in the experience that connects theoretical
field of AI and ensure the technology concepts and practical applications.
is developed safely and responsibly.
Unfortunately, some groups of people Informed by AI experts
n Teacher Dave Cross uses Experience face barriers to learning about AI, which A group of 15 research scientists
AI resources with his students means they are underrepresented in the and engineers at Google DeepMind
at North Liverpool Academy
field. By increasing diversity in AI, we bring contributed to developing the lessons. From

30 helloworld.cc
drafting definitions for key concepts, to n Embed ethics and responsibility. Key
brainstorming interesting research areas to concepts in AI ethics and responsibility
highlight, and even featuring in the videos are woven into each lesson and
included in the lessons, the group played progressively built upon. Students are
a key role in shaping the programme in introduced to concepts like data bias,
close collaboration with the Foundation’s user-focused approaches, model cards,
educators and education researchers. To and how we can use AI for social good.
bring AI concepts to life, the lessons include n Ensure cultural relevance and inclusion.
interactive activities as well as real-life Experience AI was designed with diverse
examples. One exciting project involved learners in mind and includes a variety of
Google DeepMind collaborating with activities to enable young people to pick
ecologists and conservationists to develop topics that most interest them. AIMEE WELCH
machine learning methods to study the Aimee leads Google DeepMind’s work
behaviour of an entire animal community What teachers say supporting school students and teachers to
in the Serengeti National Park and Grumeti To date, we estimate the resources have broaden participation in science and address
Reserve in Tanzania (helloworld.cc/ reached more than 130,000 students in the barriers leading to underrepresentation
MLproject). UK and beyond. We’re thrilled to hear from in the AI ecosystem. Aimee has ten years
Google DeepMind Research Scientist teachers already using the resources about of experience developing and delivering
Petar Veličković, a member of the working their impact in the classroom. Mrs J Green education programmes to increase diversity
group, shares: “AI is a technology that from Waldegrave School in London, UK, and equity in AI and STEM fields.
is going to impact us all, and therefore says: “I thought that the lessons covered
educating young people on how to interact
with this technology is likely going to be a
STUDENTS NOW FEEL IMMENSELY
core part of school education going forward.
The project was eye-opening and humbling
for me as I learned of the challenges
“ MORE KNOWLEDGEABLE ABOUT THE
associated with making such a complex topic IMPACT THAT AI HAS IN THEIR LIVES
accessible — not only to every pupil, but also
to every teacher! Observing the thoughtful
approach undertaken by the Raspberry Pi a really important topic. Giving the pupils students now feel immensely more
Foundation left me deeply impressed. I’m an understanding of what AI is and how it knowledgeable about the importance and
taking home many useful ideas that I hope works will become increasingly important place that AI has in their wider lives. These
to incorporate in my own AI teaching efforts as it becomes more ubiquitous in all areas lessons and activities are engaging and
going forward.” of society. The lessons that we trialled took accessible to students and educators alike,
The lessons have been carefully some of the ‘magic’ out of AI and started to whatever their specialism may be.”
developed to: give the students an understanding that AI We hope the programme instils
is only as good as the data that is used to confidence in both teachers and
Follow a clear learning journey,
n  build it. It also started some really interesting students, helping to address some of the
underpinned by the SEAME framework discussions with the students around areas critical school-level barriers leading to
(helloworld.cc/SEAME), which guides such as bias.” underrepresentation in AI and playing a
learners sequentially through key At North Liverpool Academy in the UK, role in building a stronger, more inclusive AI
concepts and acts as a progression teacher Dave Cross tells us: “AI is such a community where everyone can participate
framework. current and relevant topic in society that irrespective of their background.
Build foundational knowledge and
n  [these lessons] will enable Key Stage Today’s young people are tomorrow’s
provide support for teachers. Focus 3 computing students [ages 11–14] to leaders — and as such, educating and
on teacher training (helloworld.cc/ gain a solid foundation in something that inspiring them about AI is valuable for
introtoMLandAI) and support is at the will become more prevalent within the everybody.
core of the programme (watch this curriculum, and wider subjects too as more Teachers can visit experience-ai.org to
space for details on a new webinar sectors adopt AI and machine learning download all of the Experience AI resources
series being launched soon!). as standard. Our Key Stage 3 computing for free.

helloworld.cc 31
 FEATURE

© Karen Roach/stock.adobe.com
n Technology companies like to

AI ISN’T JUST ROBOTS

anthropomorphise their products to
make them seem less scary

Sway Grantham offers her tips on how to talk to young children about artificial
intelligence and avoid introducing alternate conceptions

Y
oung children have a unique and this is a concern when the media is scary, and makes you more likely to include
perspective on the world they live full of big, scary headlines about artificial them in your lives. However, when you think
in. They often seem oblivious to what’s intelligence (AI) taking over the world, of AI as a human with a name, who needs
going on around them, but then they will stealing jobs, and being sentient. As you to say ‘please’ or is ‘there to help you’,
ask a question that makes you realise they teachers, you don’t need to know all the you start to make presumptions about how it
did get some insight from a news story gory details about AI to answer learners’ works, what it ‘knows’, and its morality. This
or a conversation they overheard. This questions, but you can avoid accidentally changes what we ask, how much we trust
happened to me with a class of ten-year- introducing alternate conceptions. This the responses, and our behaviours when
olds when one boy asked, with complete article offers some top tips to help you point using the device. The device, though, does
sincerity and curiosity, “And is that when those inquisitive minds in the right direction. not ‘see’ or ‘know’ anything; instead, it uses
the zombie apocalypse happened?” He lots of data to make predictions. Think of word
had unknowingly conflated the plague AI is not a person association: if I say ‘bread’, I predict that a lot
with television depictions of zombies Technology companies like to of people in the UK will think ‘butter’. Here,
taking over the world. anthropomorphise their products and give I’ve used the data I’ve collected from years of
Absorbing media and assimilating it into them friendly names. Why? Because it makes living in this country to predict a reasonable
your existing knowledge is a challenge, their devices seem more endearing and less answer. This is all AI devices are doing.

32 helloworld.cc
 When talking to young children about AI. When learners are using streaming
AI, try to avoid using pronouns such as services for music or TV and the service
‘she’ or ‘he’. Where possible, avoid giving predicts something that they might want to
devices human names, and instead call watch or listen to next, point out that this is
them ‘computer’, to reinforce the idea that using AI technology. When they see their
humans and computers are very different. parents planning a route using a satnav,
Let’s imagine that a child in your class says, explain that the satnav system uses data
“Alexa told me a joke at the weekend — and AI to plan the best route. Even better
she’s funny!” You could respond, “I love using than just calling out uses of AI, try to have
computers to find new jokes! What was it?” conversations about when things go wrong
This is just a micro-conversation, but with it, and AI systems suggest silly options. This
you are helping to surreptitiously challenge is a great way to build critical thinking
the child’s perception of Alexa and the role around the use of computers. AI systems
of AI in it. Another good approach is to don’t always know best, because they’re SWAY GRANTHAM
remember to keep your emotions separate just making predictions, and predictions can Sway is a senior learning manager at
from computers, so as not to give them always be wrong! the Raspberry Pi Foundation, where she
human-like characteristics: don’t say that leads the educator development team
the computer ‘hates’ you, or is ‘deliberately AI and humans developing resources and training for
ignoring’ you, and remember that it’s only complement each other computing teachers around the world
‘helpful’ because it was told to be! Language There’s a delicate balance between (@SwayGrantham).
is important, and we must continually acknowledging the limitations of AI and
practise avoiding anthropomorphism! portraying it as a problematic tool that we
To help children understand how AI and
humans complement each other, ask pupils

CAN ALEXA TELL YOU A JOKE THAT MAKES the question, ‘What can’t a computer do?’

“ YOU LAUGH? YES — BUT A HUMAN CREATED

Answers that I have received before include,
‘Give me a hug’, ‘Make me laugh’, and ‘Paint

THAT JOKE. HUMANS ARE STILL REQUIRED a picture’, and these are all true. Can Alexa
tell you a joke that makes you laugh? Yes —
but a human created that joke. The computer
is just the way in which it is being shared.
AI isn’t just robots shouldn’t use. AI offers us great opportunities Even with AI ‘creating’ new artwork, it is
(actually, it rarely is) to improve the way we work, to get us started really only using data from something that
The media plays a huge role in what we on a creative project, or to complete mundane someone else created. Humans are required.
imagine when we talk about AI. For the tasks. However, it is just a tool, and tools Overall, we must remember that young
media, the challenge is how to make lines complement the range of skills that humans children are part of a world that uses AI,
of code and data inside a computer look already have. For example, if you gave and that it is likely to be ever more present
exciting and recognisable to their audiences. an AI system the prompt, ‘Write a setting in the future. We need to ensure that they
The answer? Robots! When learners hear description using these four phrases: dark, know how to use AI responsibly, by
about AI taking over the world, it’s easy for scary, forest, fairy tale’, the first output from minimising alternate conceptions. With our
them to imagine robots like those you’d find the tool probably wouldn’t make much sense. youngest learners, this means taking care
in a Marvel movie. Yet the majority of AI As a human, though, you’d have to do far less with the language you choose and the
exists within systems they’re already aware work to edit the output than if you had had examples you use, and explaining its role as
of and are using — you might just need to to write the setting description from scratch. a tool. These simple approaches are the first
help draw their attention to it. Now, say you had the perfect example of a steps to empowering children to go on to
For example, when using a word setting description, but you wanted 29 more harness this technology, and they pave the
processor, you can highlight to learners examples, so that each learner in your class way for you to simply introduce the core
that the software sometimes predicts had a different version. This is where AI can concepts of AI in later computing lessons
what word you want to type next, and that help: completing a repetitive and boring task, without first having to untangle a web of
this is an example of the computer using and saving time for humans. alternate conceptions.

helloworld.cc 33
 FEATURE

ALEXA, WHAT ARE

YOU THINKING ABOUT?
Judy Robertson describes recent research into what children already know
about AI based on their interactions with smart speakers at home

T
he grandfather of neural networks, and Google Home. At the University examples of AI, rather than being literally
Geoffrey Hinton, recently of Edinburgh, we recently researched human, although about two-thirds of them
commented that he didn’t think it would children’s attitudes towards smart speakers, thought that they might think like a human.
be long before artificial intelligence (AI) to find out what they already know about This research showed that children aren’t
became more intelligent than humans. AI (helloworld.cc/Robertson) and help us sure what to think of smart speakers. In the
Not everyone agrees, but AI is certainly develop educational materials for AI literacy. survey, when asked whether smart speakers
advancing eye-wateringly fast. I believe have feelings, or can make decisions for
that, as educators, we have a responsibility Children’s views on smart speakers themselves, many answered ‘maybe’. When
to teach young people about how AI We conducted a survey of 166 eight- to posed this question in the focus groups, the
technology works and help them think eleven-year-olds, with follow-up focus younger children sometimes just admitted
about how it should be used in society. groups. Of the children, 93 percent had a they didn’t know. The older children debated
We must remember, though, that AI isn’t smart speaker at home, mostly Amazon’s the question together, backing up their
something that is just about to happen in Alexa. The children generally thought that answers with examples of what Alexa does
children’s lives. It has been part of their devices such as Alexa were smart, often when they interact with it at home. For
home lives for at least five years, in the cleverer than themselves, but less clever than example, one child reported, “I said ‘Arrgg,
form of smart speakers such as Alexa scientists. They knew that the devices were just play the song already.’ He [Alexa] was

n AI is already part of children’s home

lives in the form of smart speakers
© shara/stock.adobe.com

34 helloworld.cc
 had inaccurate ideas about what the smart programming had misunderstandings —
speakers would do with their information. some thought that smart speakers were
One child speculated, “If you tell Alexa programmed in advance with an answer to
secrets, I feel those messages travel to every possible question that a user might
Alexas throughout the Alexa Amazon pose. There is a gap in children’s knowledge
network, and then if someone’s telling their here, even if they pay close attention to their
secrets to Alexa and someone else asks computational thinking lessons; thinking of
them to tell their secrets but then it tells machine learning techniques used in smart
YOUR secrets!” This scenario is improbable speakers as simply procedural algorithms
if you know that Alexa devices are designed does not capture the insight that they are
to be associated with specific user predictive statistical models.
accounts, and that they are not networked
JUDY ROBERTSON together in such a way that information can Suggestions for teachers
Judy is a professor of digital learning at the inadvertently leak between devices. If you So what can educators do to help children
University of Edinburgh. She researches are making an analogy with the way that understand AI better? A good starting
computer science education, data literacy, gossip travels around a school playground, place is to talk to children about how the
and AI literacy. She also designs technology though, it seems more plausible. In this AI technology they use in their everyday
with and for children. case, the analogy with human interactions lives works (helloworld.cc/AI-literacy).
raises a threat that is not likely to happen. As part of this discussion, inform children
The children were horrified by the idea of about privacy settings for the devices they
like, ‘Don’t be rude, OK?’ So I don’t know if the other people in their house being able commonly use, discuss with them what
he has feelings.” to see records of their interactions with their settings they feel comfortable with, and
It’s not surprising that children are smart speakers, and told us they would help them to put the risks into perspective.
puzzled about whether smart speakers have
feelings, because they are often designed to
CHILDREN ARE OFTEN PUZZLED ABOUT
“
behave as if they do. When the device sings
Happy Birthday to a child or thanks them
for a compliment, it builds up the impression WHETHER SMART SPEAKERS HAVE FEELINGS
that it is humanlike. So, the next time the
device fails to respond when someone
shouts at it, the child may reasonably guess change their behaviour if that were the case. Teachers can also help to explain how
that it is sulking. If the child doesn’t have any They did not understand that, in fact, Alexa machine learning works, and this will
technical understanding of how the device does keep a transcript of all interactions probably work best with a class that already
works, their only method for making sense with all users in the house, which can be understands the concept of simple algorithms
of it is to reason about its behaviour as they accessed by the account holder (often the (helloworld.cc/algorithms). Machine learning
would reason about another person. parents or carers). The potential harm to algorithms enable the computer to figure out
the child of a privacy invasion by another how to do tasks without the programmer
Misunderstandings about family member is possibly more immediately explicitly writing down rules. The book
smart speakers bothersome than the misuse of data by a Neural Networks for Babies by Chris Ferrie
An important part of our study was the goal third-party company. and Dr Sarah Kaiser is an excellent picture
of understanding the misconceptions young Another set of misunderstandings about book which introduces the main concepts of
people have about smart speakers and smart speakers was related to how they give machine learning to a young audience.
AI more broadly. The problem with trying such apparently intelligent answers. Some Just like adults, some children may be
to make sense of how a smart speaker children thought that a human intercepted excited by new AI technologies, and some
behaves as if it were human is that it leads the questions from users and thought of the may be fearful of the idea that AI could
to inaccurate predictions. In the study, we responses. Memorably, one child surmised replace humans. As teachers, we can help
found that the children both overestimated that a “really smart guy plugged himself in” them make sense of the possible future of
and underestimated the capabilities of the to Alexa to download his intelligence and AI by beginning with what they already
technology. Privacy mattered a lot to the build an army of Alexas. Even the children know about the AI they commonly
children, for example, but they sometimes who had been taught about computer encounter at home.

helloworld.cc 35
 FEATURE

AI MADE EASY:
TEACHING AI TO KS2 AND KS3
Rebecca Mack shares her experience of creating and delivering a new curriculum
topic about artificial intelligence in both primary and secondary settings

Implementation and student voice

A
rtificial intelligence (AI) has gained Teaching AI to students is crucial due
significant attention worldwide to its rapid development and integration The implementation of the AI curriculum
recently. It’s even a prominent topic among into various aspects of our lives. Educating involved engaging activities and discussions
young students who are now familiar students about AI empowers them to throughout. Here are some specific topics
with, and sometimes actively using, AI actively participate in the digital age, covered and the students’ feedback.
tools like ChatGPT. As educators, we comprehend how AI functions, explore its
need to prepare students for the future potential applications, and think about the n Machine learning: we used the Hour of

by equipping them with knowledge ethical considerations involved. Integrating Code AI for Oceans tool (code.org/oceans)
and understanding of AI’s workings practical AI tools like ChatGPT into the to introduce ML. Students explored AI
and limitations. This article explores the curriculum offers students an immersive learning steps while addressing real-
journey of designing and implementing learning experience. world environmental challenges. Their
an AI curriculum for Year 8 students, Our AI curriculum was designed with the feedback reflected enthusiasm and
aged 12–13, at a preparatory school. inclusion of AI tools; as a computing teacher, improved understanding, with one student
It emphasises the importance of AI I used ChatGPT to show my students its mentioning how the tool sparked their
education in shaping digital literacy in the functionalities and foster a collaborative interest and enhanced their knowledge of
next generation. learning environment where they could work the principles.
together to discover its potential.
I used ChatGPT as a tool to help form n Natural language processing and virtual

ideas for what to cover in our curriculum. personal assistants: we examined adverts
It helped me to identify key areas, for popular voice assistants like Siri, Alexa,
such as machine learning (ML), natural and Google Assistant to explore NLP and
language processing (NLP) and virtual VPAs. Analysing these examples provided
personal assistants (VPAs), text-to-image insights into how NLP enables machines
generators, deepfakes, and the crucial to understand and respond to human
topic of bias and ethics in AI. By exploring language. Students tried out a variety
these areas, our students gained a better of chatbots, like Octa.ai, before creating
understanding of AI’s core components and and acting out their own dialogue trees,
its real-world implications. promoting critical thinking about NLP
REBECCA MACK To assess their learning, the students capabilities and limitations.
Rebecca is the head of computing and completed a short quiz and gave their
digital learning at Bilton Grange Preparatory opinions on the topic four weeks after n Text-to-image generators: text-to-image

School in the UK. Rebecca holds a master’s completion. This assessment also helped generators emerged as a popular and
degree in physics and another in education to evaluate the curriculum’s effectiveness enjoyable topic among students. Activities
(@Becca_Mack). and gather feedback to improve its involved determining whether images were
future iterations. real or fake, writing stories using Mad Libs

36 helloworld.cc
templates, and using Dream by WOMBO After teaching the AI curriculum to Year 8
(dream.ai) to illustrate their narratives.
These projects sparked imagination and
students (aged 12–13), I adapted it for Year
5 students (aged 9–10) with great success. STUDENT
provided tangible demonstrations of AI’s While the core content remained the same, FEEDBACK
ability to generate images from text inputs. I tailored the delivery to suit a younger
audience. For instance, during discussions on ”I think it was interesting to see how AI is
n Deepfakes: we began exploring ethics, I provided more support and guidance, programmed to determine whether things
deepfakes by watching the UK’s Channel focusing on understanding how others are classified as ‘trash’ or ‘sea animals’. It
4 video apparently featuring the late might perceive and feel about AI-related made me understand the collaborative nature
Queen’s famous TikTok dance. This initiated topics, taking a one-step-removed approach. of AI and how it learns with human help.”
discussions on the ethics of deepfakes, Also, in our image generation lesson, I used
emphasising critical thinking and media ChatGPT to collaboratively create a story “Learning about deepfakes helped
literacy. The topic also connected with with the students before they used an image us understand what is fake on the
e-safety discussions, highlighting the need generator to illustrate it. I also used this task internet and how we can avoid it.”
for caution when encountering online to integrate elements of their English lessons,
content. We also tied in ProjectEVOLVE’s such as verbs and adjectives. The students
‘Managing online information’ lesson that didn’t directly interact with the AI but worked AI technology across the school, recognising
explores fake news (projectevolve.co.uk). alongside me as we explored ChatGPT’s the value of AI education in preparing
students for the future. If we are using
the technology to support our teaching
I PREDICT THAT SOON EVERY COMPUTER
“
and learning, it is key for our students to
understand it too.
SCIENCE CURRICULUM WILL EMBED AI In conclusion, the positive feedback from
CONCEPTS THROUGH ALL KEY STAGES students, quiz results, and the enthusiasm
of the senior leadership team demonstrate
the success of our AI curriculum in engaging
n Ethics and bias: this topic resonated capabilities. The Year 5 students not only students and fostering their understanding
strongly with students, and despite ranking grasped the material but also showed a level of AI concepts. Our next steps are to
lower in terms of enjoyment in the final of enthusiasm and engagement that rivalled continue the collaboration between
quiz, it was voted the most useful topic that of their Year 8 counterparts! educators and school leadership to ensure
for the future. We discussed the ethical Our school’s senior leadership team has AI technology is effectively integrated
considerations of self-driving cars, and shown a keen interest in exploring AI from across various subjects, equipping students
revisited the issue of bias in AI systems, a teaching and learning perspective. We with the necessary skills and knowledge to
emphasising the importance of addressing are currently collaborating to incorporate thrive in the digital age.
and mitigating bias for fairness and
inclusivity. We also touched on the concept
© Marta Sher/stock.adobe.com

of the Turing test here — several students

worked further on it at home!

Reflections and the senior

leadership team’s perspective
The end-of-topic quiz resulted in an
average score of 85 percent. After four
weeks, we repeated the quiz and saw a
slight decrease to 79 percent. As expected,
this demonstrates the need for continuous
reinforcement. Based on the speed at which
n Discussing the role AI
AI is developing, I predict that soon every technology plays in fake news and
deepfakes emphasises critical
computer science curriculum will embed AI
thinking and media literacy
concepts throughout all the Key Stages.

helloworld.cc 37
 FEATURE

© BBC / Better Images of AI / Quantified Human / CC-BY 4.0

PICTURING AI
How do we represent artificial intelligence in images that are helpful?

H
ello World is accustomed to But how do you make a cover that applications are built to answer questions or
visually representing concepts communicates all that to readers in a single help diagnose illnesses. Other AI applications
that are hard to depict. In this issue, we image that is shorthand for the theme of could be built for harmful purposes, such as
wanted to represent the many facets of the issue, while not giving in to the current spreading fake news. AI applications do not
teaching about AI technology — the social, tropes? If you search online for images think. AI applications are built to carry out
ethical, creative, gender-balance issuews, around artificial intelligence, they usually tasks in a way that appears to be intelligent.”
as well as the potential for AI technology feature robots, neural connections in the With this explanation in mind, why are
to enhance our lives. We also wanted a shape of a human brain, or a version of the majority of current AI images unhelpful?
human artist, who we could exchange Michelangelo’s The Creation of Adam with a What would be a better type of image?
ideas with, to illustrate it. human hand bringing life to a robot’s hand.
Although these images instantly Better images of AI
communicate ‘AI’ to a broad range of people While creating this magazine, we kept
because they reference well-known sci-fi referring back to the guidelines from
visions of the technology, they are unhelpful the Better Images of AI organisation
as shorthand for what existing AI actually is. (betterimagesofai.org). The group is a
collaboration between various global
What exactly is AI? academics, artists, diversity advocates,
The glossary to our Experience AI and non-profit organisations. It aims to help
(experience-ai.org) programme explains create a more representative and realistic
AI like this: “Artificial intelligence (AI) is the visual language for AI systems, themes,
design and study of systems that appear applications, and impacts.
to mimic intelligent behaviour. Some AI We applied the following guidance,
applications are based on rules. More often from the blog Better Images of AI: A Guide
MEG WANG now, AI applications are built using machine for Users and Creators (helloworld.cc/
Meg is the editor of Hello World. learning that is said to ‘learn’ from examples imageguide), by Dr Kanta Dihal and Tania
in the form of data. For example, some AI Duarte, to our image choices.

38 helloworld.cc
Quantified Human by Alan Warburton
This image, from Better Images of AI, shows
a young Black man wearing a black coat AI IMAGE GUIDANCE
staring past the camera in front of a blue
cloudy sky. The scene is refracted in different Avoid using Instead use
ways by a fragmented glass grid. This grid is
n The colour blue n Honesty: showing accurately what the technology
a visual metaphor for the way in which new
n Descending code can do and nothing more
AI and machine learning technologies can
n The human brain n Humanity: not depicting AI as humanlike, but
be used to extract and analyse behavioural
n Science fiction references showing people working with AI technology
and demographic data in innovative ways.
n White robots n Necessity: not using images of AI when the story
Some of the grid squares reveal graphical
n White men in suits they illustrate is not about AI
interpretations of the man that exceed the
n Anthropomorphism n Specificity: using images that are specific to the
capabilities of human vision, indicating how
n Variations on The Creation of Adam type of technology or impact being discussed
cutting-edge technologies offer ways to
augment traditional human understandings of
complex phenomena. A neural network diagram
is overlaid, familiarising the viewer with the
FUTURISTIC IMAGES CREATE UNHELPFUL
“
formal architecture of AI systems.

STEREOTYPES ABOUT AI TECHNOLOGY

and programmed to mimic intelligent come from stock image libraries, and tend
behaviour which still needs human input. to perpetuate the same old AI tropes rather
than capturing AI’s many applications. The
n White robots and white men in suits Better Images of AI image library showcases
Metaphors to avoid, and why Lack of representation and lack of some alternative approaches. The
n Robots — either scary or friendly diversity amplifies inequalities. The organisation hopes to inspire users, creators,
Inaccurate images of AI can be ongoing issue of bias extends to and commissioners of stock images to think
intimidating and confusing to people the visual representation of AI as more about what they are communicating
who don’t normally discuss or think well. It is challenging to find images and how this can be more authentically,
about AI. Terminator-like images of AI that include representation of inclusively, and creatively represented.
perpetuate misconceptions that AI is people of colour, women, and other To return to how we approached
disturbing and dystopian, something to underrepresented groups. representing AI visually in Hello World,
mistrust and worry about. Neither do you’ll see that there is only one image of
we want to anthropomorphise AI as a n Descending code or futuristic robots in this issue. We stuck with the
friendly robot (see pages 26–27). AI is neural networks Better Images of AI guidelines of showing
technology that we already use in our Futuristic images create unhelpful accurately what technology can do and
everyday lives, in the form of stereotypes about AI technology, and a nothing more.
voice-activated assistants (such as Siri lack of realism stops people recognising We still have to bridge the gap between
and Google) or satnav, and it rarely where AI already exists in their lives. better images of AI and the general public’s
takes the form of a robot. Realistic images of AI — whether they be visual shorthand for AI. When planning this
recommendations on Spotify or Netflix, issue’s magazine cover, we discussed using
n Human brain or The Creation of Adam or data annotators who label data so it images of computer vision or music
Exaggerating the capabilities of AI can be recognised by a machine learning generators, but these are symbols that are
systems sets the wrong expectations. model — can help people understand how mainly recognisable by those who are
This imagery oversells the scope of their day-to-day lives already include AI. already familiar with AI. We also needed an
what the technology is capable of, image that immediately grabbed the
creating a disconnect between how What would be a good image of AI? reader’s attention, as a good cover should.
AI is viewed by wider audiences and Better Images of AI have created a free I’m not sure this cover captures everything
the reality. AI is not a sentient being. visual library to help us build better we wanted to convey about AI, but I think
It’s a system that’s been designed representations of AI. Most images of AI it’s a move towards better AI imagery.

helloworld.cc 39
 FEATURE

CREATING AN AI GLOSSARY
Explaining AI terms for young people and educators

W
hat do we talk about when we talk these terms continue to evolve, we prefer develop their understanding. If we use it
about artificial intelligence (AI)? ‘explanation’ to ‘definition’. The explanations inconsistently, this can lead to alternate
The term ‘AI’ is used to describe so many are useful because: conceptions (misconceptions) that can
different things nowadays, it’s difficult 1. They ensure that we give learners interfere with learners’ understanding. You
to know straight away what anyone and teachers a consistent and clear can read more about this in our Pedagogy
means when they use the term. However, understanding of the key terms across Quick Read on alternate conceptions
without a shared understanding of what all our resources. Within the Experience (helloworld.cc/alternateconceptions).
AI and related terms mean, we can’t talk AI lessons for Key Stage 3 (ages 11–14), Among our principles for writing
about them or educate young people these key terms are also correlated to the explanations of AI terms were that the
about the field. target concepts and learning objectives explanations needed to:
When we started designing materials presented in the learning graph. n Be accurate
for the Experience AI learning programme, 2. They help us talk about AI, and AI n Be grounded in education research
we decided to create short explanations of education, within our team. With a best practice
key AI and machine learning (ML) terms. As shared understanding of what terms n Be suitable for our target audience (Key
such as ‘AI’, ‘ML’, ‘model’, or ‘training’ Stage 3 learners, 11- to 14-year-olds)
actually mean, our conversations are n Be free of terms that have alternative
n We spent months thinking about, writing, correcting,
discussing, and justifying the glossary explanations
more productive. meanings in computer science, such
You can find 32 explanations in the glossary as ‘algorithm’
that forms part of the Experience AI lessons We engaged in an iterative process of
© Teresa Berndtsson / Better Images of AI / CC-BY 4.0

(experience-ai.org). Here’s an insight into writing explanations, gathering feedback

how we arrived at the explanations. from our team and our Experience AI
project partners at Google DeepMind, and
Reliable sources then adapting the explanations. We went
In order to ensure that the explanations were through the feedback and adaptation cycle
as precise as possible, we first identified until we all agreed that the explanations met
reliable sources. These included, among our principles.
many others: An important part of what emerged,
n The Oxford English Dictionary (oed.com) aside from the explanations of AI terms
n Google’s machine learning glossary themselves, was a blueprint for how not to
(helloworld.cc/MLglossary) talk about AI. One aspect of this is avoiding
n The Alan Turing Institute’s data science anthropomorphism, which Ben Garside
and AI glossary (helloworld.cc/ writes about on pages 26–27 of this issue.
Turingglossary) As part of designing Experience AI,
n Well-recognised AI courses, such creating the explanations helped us to:
as Andrew Ng’s AI for Everyone n Decide which technical details we
(helloworld.cc/Ngglossary) needed to include when introducing AI
n Articles included in the AITopics concepts in the lessons
publication of the AAAI (aitopics.org) n Figure out how to best present these
technical details
Principles for explaining AI terms n Settle debates about where it would be
Vocabulary is an important part of teaching appropriate, given our understanding
and learning. When we use vocabulary and our learners’ age group, to abstract
correctly, we can support learners to or leave out details

40 helloworld.cc
n Our ‘artificial intelligence’ explanation, broken up into the parts
of the semantic wave. Red = top of the wave; yellow = middle of n Artificial intelligence (AI) is the design
the wave; green = bottom of the wave
and study of systems that appear to
mimic intelligent behaviour (top of the
Using education research to explain wave)
AI terms n Some AI applications are based on rules; VERONICA CUCUIAT
One of the ways in which education more often now, AI applications are built Veronica is a research scientist who is
research informed the explanations was using machine learning that is said to interested in how young learners engage with
that we used semantic waves (helloworld. ‘learn’ from examples in the form of data digital creative ways. She’s also a mum and a
cc/semanticwaves) to structure each term’s (middle of the wave) cycling enthusiast, and in her spare time you
explanation in three parts: n For example, some AI applications can find her at the allotment with her chickens.
1. Top of the wave: the first one or two are built to answer questions or help
sentences are a high-level abstract diagnose illnesses; other AI applications
explanation of the term, kept as short as could be built for harmful purposes, quite a few inconsistencies with our
possible, while introducing key words such as spreading fake news (bottom of principles, as well as a couple of technical
and concepts. the wave) inaccuracies. Perhaps I could have tweaked
2. Bottom of the wave: the middle part of n AI applications do not think (middle of the prompt for the chatbot in order to get a
the explanation unpacks the meaning the wave) better result. However, relying on a chatbot’s
output would have meant missing out on
some of the value of doing the work of
SOME OF THE EXPLANATIONS WENT
“
writing the explanations in collaboration with
my team and our partners.
THROUGH TEN OR MORE ITERATIONS The visible result of that work is the
explanations themselves. The invisible
result is the knowledge we all gained, and
of the term using a common example, n AI applications are built to carry out tasks the coherence we achieved as a team, both
in a context that’s familiar to a young in a way that appears to be intelligent of which enabled us to create high-quality
audience. (top of the wave) resources for Experience AI. We wouldn’t
3. Top of the wave: the final one or two have gotten to know what resources
sentences repack what was explained Was it worth our time? we wanted to write without writing the
in the example in a more abstract way Some of the explanations went through ten explanations ourselves and improving them
again, to reconnect with the term. The or more iterations before we agreed that over and over. So yes, it was worth our time.
end part should be a repeat of the top they were suitable for publication. After The process of creating and iterating the
of the wave at the beginning of the months of thinking about, writing, correcting, AI explanations highlights how opaque the
explanation. It should also add further discussing, and justifying the explanations, field of AI still is, and how little we yet know
information to lead to another concept. it’s tempting to wonder whether I should about how best to teach and learn about it.
Most explanations also contain ‘middle of have just prompted an AI chatbot to At the Raspberry Pi Foundation, we now
the wave’ sentences, which add additional generate the explanations for me. know just a bit more about that, and are
abstract content, bridging the ‘bottom of the I tested this idea by getting a chatbot excited to share the results with teachers
wave’ concrete example to the ‘top of the to generate an explanation of ‘artificial and young people. The glossary of AI
wave’ abstract content. intelligence’ using the prompt, “Explain what explanations is in its first published version,
Here’s our ‘artificial intelligence’ artificial intelligence is, using vocabulary and we will continue to improve it as we find
explanation, broken up into the parts of the suitable for KS3 students, avoiding out more about how to best support young
semantic wave: anthropomorphism.” The result included people to learn about this field.

helloworld.cc 41
 FEATURE

TEACHING AI EXPLAINABILITY:
EMPOWERING DISCERNING
CONSUMERS OF AI TECHNOLOGIES
Mac Bowley shares the ways developers are trying to make more transparent AI models
and the importance of teaching students to be critical consumers of AI technology

I
n the rapidly evolving digital code is being generated and used by others could not detect their faces as effectively
landscape, students are without their knowledge or consent. If you as their lighter-skinned friends and
increasingly interacting with AI-powered came across an AI-generated artwork that family. Explainability allows us not only to
applications when listening to music, features a face resembling yours, it’s natural understand but also to challenge the outputs
writing assignments, and shopping online. to want to understand how a photo of you of a model if they are found to be unfair.
As educators, it’s our responsibility to was incorporated into the training data. In essence, explainability is about
equip them with the skills to critically There will also be instances where a accountability, transparency, and fairness,
evaluate these technologies. model seems to be working for some people which are vital lessons for children as they
A key aspect of this is understanding but is inaccurate for a certain demographic grow up in an increasingly digital world.
‘explainability’ in AI and machine learning of users. This happened with Twitter’s
(ML) systems. The explainability of a model (now X’s) face detection model in photos; Routes to explainability
is how easy it is to ‘explain’ how a particular the model didn’t work as well for people Some models, like decision trees, regression
output was generated. Imagine having a job with darker skin tones, who found that it curves, and clustering, have an in-built level
application rejected by an AI model, or facial
recognition technology failing to recognise
you — you would want to know why.
Establishing standards for explainability is
TEACHING ABOUT MODEL CARDS
crucial. Otherwise we risk creating a world
You can use classroom discussions to explore what different stakeholders might want to know about a model
where decisions impacting our lives are made
and how it could be included on a model card (see next page for more on model cards). Here are three example
by opaque systems we don’t understand.
discussion topics with stakeholders for you to consider with your students.
Learning about explainability is crucial for
students to develop digital literacy, enabling Medical diagnosis If a model is used to decide on a course of Stakeholders: doctors,
them to navigate the digital world with treatment for a patient, what would each patients, and family
informed awareness and critical thinking. stakeholder need to know about the model to members
feel comfortable following the advice?
Why explainability is important
AI models can have a significant impact on Text generation When evaluating the use of a text generation Stakeholders: authors,
people’s lives in various ways. For instance, (like ChatGPT) model in their work, what would the stakeholders students, and teachers
if a model determines a child’s exam need to consider about the model’s purpose
results, parents and teachers would want to and accuracy?
understand the reasoning behind it.
Artists might want to know if their Music generation If an AI model was used to generate music, what Stakeholders: artists,
creative works have been used to train a questions would each of the stakeholders have listeners, and a record

model and could be at risk of plagiarism. about the data used to train it? company

Likewise, coders will want to know if their

42 helloworld.cc
 you buy an item of food in a supermarket,

FIND OUT MORE you can look at the packaging and find all
sorts of nutritional information, such as the
ingredients, macronutrients, allergens they
The AI team at Google introduced model may contain, and recommended serving
cards as a first step in encouraging
sizes. This information is there to help
transparency from model developers. Read
inform consumers about the choices
more in this research paper (helloworld.cc/
modelcardresearch). they are making. n A model card mock-up from the Experience AI lessons
Model cards attempt to do the same
Along with the research, Google have
thing for ML models, providing essential and the consequences they might have
released two examples of model cards you
can use with your students (helloworld.cc/
information to developers and users of a for different groups of people. By teaching
Googlemodelcards). model so they can make informed choices students about explainability, we are not
about whether or not they want to use it. only educating them about the workings of
Model cards include details such as the these technologies, but also teaching them
of explainability. There is a visual way to developer of the model, the training data to expect transparency as they grow to be
represent these models, so we can pretty used, the accuracy across diverse groups of future consumers or even developers of AI
accurately follow the logic implemented by people, and any limitations the developers technology.
the model to arrive at a particular output. uncovered in testing. Most importantly, model cards should be
A decision tree works like a flowchart, A real-world example of a model card is accessible to as many people as possible
and you can follow the conditions used to Google’s Face Detection model card. This — taking this information and presenting it
arrive at a prediction. Regression curves details the model’s purpose, architecture, in a clear and understandable way. Model
can be shown on a graph to understand performance across various demographics, cards are a great way for you to show your
why a particular piece of data was treated and any known limitations of their model. students what information is important for
the way it was, although this wouldn’t give This information helps developers who people to know about a model and why
us insight into exactly why the curve was might want to use the model to assess they might want to know it. Model cards
placed at that point. Clustering is a way of whether it is fit for their purpose. can help students understand the
collecting similar pieces of data together importance of transparency and
to create groups (or clusters) with which Transparency and accountability accountability in AI.
we can interrogate the model to determine As the world settles into the new reality of
which characteristics were used to create having the amazing power of AI models at
the groupings. our disposal for almost any task, we must
However, the more powerful the model, teach young people about the importance
the less explainable it tends to be. Neural of transparency and responsibility.
networks, for instance, are notoriously hard As a society, we need to have hard
to understand — even for their developers. discussions about where and when we
The networks used to generate images or are comfortable implementing models
text can contain millions of nodes spread
across thousands of layers. Trying to work
out what any individual node or layer is
doing to the data is extremely difficult.
Regardless of the complexity, it is still
vital that developers find a way of providing MAC BOWLEY
essential information to anyone looking Mac is a computing educator who
to use their models in an application or has worked in almost every context
to a consumer who might be negatively imaginable: after-school clubs, holiday
impacted by the use of their model. camps, enrichment days, and teaching
GCSE students. Mac is passionate about
Model cards empowering people to use technology to
One suggested strategy to add transparency
n A decision tree that classifies animals based on their solve problems that matter to them.
to these models is using model cards. When characteristics; you can follow these models like a flowchart

helloworld.cc 43
 FEATURE

RECOMMENDER SYSTEMS:
THE HIDDEN AI STEERING OUR CHOICES
Edward Abel explores recommender systems, what they are, and how they operate,
and looks at wider considerations around their growing influence in our digital lives

E
very time we visit video websites Additionally, it keeps a log of every video a people who liked the same 200 items as
such as YouTube or TikTok, or user watches, along with information about you also liked this item, and we think you
music streaming sites like Spotify or Apple how they watched it, such as whether they will like it too’, which has been shown to be
Music, we are presented with suggested stopped before the end, watched early in a successful way to suggest content a user
content to consume. Such recommended the morning, or watched with no sound. will like. However, it can struggle to make
suggestions are invariably generated Its aim is then to use such information to good recommendations for new users, as it
by recommender (or recommendation) suggest more content to a user. But how relies on user histories to determine patterns
systems, and these systems are powered do recommender systems choose what with other users. Additionally, it can struggle
by AI tools. content to suggest? to provide meaningful explanations as to
why it makes its suggestions.
User personalisation Recommender system approaches Content-based filtering works by
Recommendation systems look to make use Two of the most common approaches for comparing items and suggesting similar
of information about the previous videos making recommendations are collaborative ones. To do this, it uses information
or songs we have consumed, together filtering and content-based filtering. that describes each item. For a movie,
with data about the characteristics of all Collaborative filtering aims to find for example, this
the available videos or songs, to suggest similarity patterns between users; two users information could
personal recommendations that it thinks who liked the same items (such as videos include its genre,
we will like. For example, for every video on on YouTube or songs on Spotify) in the past actors, running
YouTube, it documents data about the topic, will probably both like similar items in the length, and much
the length, who features in it, and so on. future. In effect, the system tells us, ‘Other more. Then, when

n AI-powered recommender systems are an increasingly

inescapable component of our digital lives
© Yulyana/stock.adobe.com

44 helloworld.cc
a user watches or likes a movie, the system systems are now considering more nuanced
can recommend other movies with similar ideas about recommendation objectives,
characteristics. Content-based filtering such as serendipity, to suggest items users
does not require lots of user history data, might like but have never considered or
making it suitable for new users. It can also heard of before. Such ideas can even be
provide more meaningful explanations; for taken further, to provide recommendations
example, you may know that it suggested that challenge a user’s views, or provide
a particular movie because it has a certain them with a chance to confront alternative
actor in it. However, due to its focus on item sides of an argument.
similarities, it can struggle to make diverse Given concerns that big tech companies
recommendations. For example, for a user may not always have users’ best interests
who has just watched Toy Story, it may at heart, there is a growing consensus that
simply recommend Toy Story 2, 3, and 4 if it good recommendations are also ones where EDWARD ABEL
deems those to be the most similar items. additional explanation is provided, detailing An assistant professor (lecturer) at the
Today, many websites use hybrid why items are being suggested. Such University of Southern Denmark, Edward
recommender systems that utilise both explainability is sought both by individuals, explores research within data science
collaborative and content-based filtering, who want to understand why certain items and AI, and has previously been involved
to leverage each system’s strengths are being suggested, and by governments in running Code Clubs and CoderDojos in
and offset their disadvantages. For and supreme courts wrestling with the United Kingdom.
example, Netflix utilises both data about regulations and accountability. However,
its videos (content) and data about its
users’ viewing histories (collaborative
THERE’S A RISK RECOMMENDATIONS MAY
“
filtering) when making recommendations.
The effectiveness of such systems
lies in their ability to leverage machine NOT ALIGN WITH WHAT USERS TRULY ENJOY
learning, and vast amounts of data, to
calculate valuable recommendations.
For example, collaborative filtering can there is some apprehension from companies showing each other what their home pages
determine underlying patterns and hard- that greater explainability may reveal too look like.
to-spot connections between millions of much, facilitating a loss of competitive As such systems become more
users, and utilise them in making good advantage, and that there is a trade-off commonplace, they are influencing more
recommendations. But what do these between performance and explainability and more of the content that we are
‘good’ recommendations look like? that is not worth the compromise. reading, watching, and listening to. This
means they have incredible power to
What are good recommendations? Influence and power influence which videos go viral, and which
Recommender systems and the AI Recommendation systems are becoming an do not. The wielding of such influence by
technologies that power them are built by increasingly inescapable component of our big tech companies raises concerns that
humans, and therefore invariably reflect, digital lives, with users today just as likely recommender systems could be used for
either explicitly or implicitly, their creators’ to discover new content via such systems dishonest purposes. A company could try to
objectives and biases. For big tech websites, as they are by searching directly for it. influence customer behaviour towards
the main goal of their recommender As a result, young learners have regular buying more expensive products, or could
systems is to suggest content that will exposure to such systems, which can even aim to influence a whole political
keep a user’s attention, and so keep them present a great opportunity to engage them campaign. As recommender system
on the site as long as possible. When via these direct links to their own lives. As algorithms get more and more complex,
companies just prioritise retaining users’ educators, we can foster awareness of such issues might even occur without the
immediate attention, there’s a risk that the recommender systems by getting learners designers intending for them to happen. It is
recommendations may not align with what to try and identify when they encounter therefore essential that young learners are
users would truly enjoy. Recently, we have them. Additionally, multiple learners who aware of recommender systems and are
begun to question the notion of what a good use the same site, such as YouTube, can curious regarding their operation, purpose,
recommender system is, and recommender observe the impacts of such systems by and intentions.

helloworld.cc 45
 FEATURE

RECOMMENDATIONS TO
DEMYSTIFY MACHINE LEARNING
James Abela outlines activities to unpack the mysteries of artificial
intelligence, focusing on recommendation systems

T
oo much has been written about recommendation, or they did not. This With this data, you can then
AI being a black box — the subset of machine learning is used on sites begin creating an AI-based snack
idea that computer scientists don’t such as Netflix, YouTube, and Amazon. recommendation system for the class,
understand how AI systems make Every time you see a recommended video initially ordered by the number of votes.
decisions. Although there is some on these sites, there is a recommendation You could teach students how to create
truth to this once a system has been system behind the scenes trying to guess the code for this via a sorting lesson (for
scaled up, the fundamentals of these the best video choice for you. example, helloworld.cc/adasorting), or by
programs are still based on binary, and Students need to understand that using the built-in Python sorting function:
on computers selecting the right choice. everything in a computer is binary, so to items.sort(reverse=True). You can try the
As computing educators, it is important start with, you can ask students for a simple code for this at helloworld.cc/reccode.
that we demystify AI for our students, thumbs up or thumbs down to each of the You now have your initial weightings, and
and unpack the idea of it being a mystical following snacks: based on these, the system will order the
unknown beyond our control. In this list of snacks. If there is a tie in the system,
article, I suggest some activities for lower n Bakewell tart most languages will default to sorting
secondary (ages 11–16) to do just that! n Crisps the items by alphabetical order. A real
n Ice cream recommendation system might randomly
It all starts with binary n Jaffa Cake choose between the tied items, or even give
A good place to start is with a machine n Pork pie a certain amount of time with each at the
learning recommendation system. Not only n Sausage roll top to see which works better.
are these a relatable hook for students, n Scone
but they also go right back to the roots n Banana Considering multiple factors
of binary: either somebody clicked on the n Shortbread The reason many recommendation systems
are seen as black boxes is because of the
large number of factors that go into the
system. In our example, we have used just
one factor so far — thumbs up or thumbs
down. The next factor will be to see who
Metrics chooses what, and then dynamically change
• Weather the ranking. This will make for a better
Input • Popularity Weightings Output recommendation system, as it will collect
• Frequency Ordered
• Cost more information about preferences from
each user.
A relatable way to exemplify this to
students is through football, where a
n Recommender model for snack choices manager chooses to play a footballer
based on the number of goals they score.
This might be an effective measure for a

46 helloworld.cc
 70 percent of a film or single episode of a any healthy snack choices. In this case, the
series) for the first 7 and 28 days on Netflix machine learning does not have to be fully
(helloworld.cc/netflix2019). autonomous, and you can teach students
If we return to our snacks, there are about programming in some key rules to
many potential factors that could affect the ensure it highlights healthy snacks.
popularity of a snack, such as: If the people creating AI systems do not
n Weather: ice cream is particularly think carefully about the factors they choose
popular on hot days and the weighting they give them, there
n Popularity: the overall popularity of the is clearly potential for dire consequences.
n A spreadsheet to show how the snack ranking is calculated snack Back in 2015, a Mercedes executive said
n Frequency ordered: people sometimes that their future autonomous cars would
striker, but even for a striker, this does not get bored of the same snacks prioritise protecting the driver of their cars
consider the passes that enabled a goal n Cost: the cheapest snack might be the at the expense of pedestrians (helloworld.
to be scored, or the number of times that one people choose more frequently cc/merc2016). An Air Force colonel similarly
they lost possession. This becomes even highlighted the negative consequences
more complex for midfielders, and from Once you have decided on the factors, of not training an AI system carefully, by
this apparently simple game, you see the you then need to add a weighting for suggesting that in order to win points
large number of complexities that would which factors are the most important, in a battle simulation, an AI drone might
need to be programmed to make effective and from there, you will get a ranking. It kill its operator if they stand in the way
recommendations and predictions. There is useful to use a spreadsheet for this (for of a successful mission (helloworld.cc/
is now a platform for scouting new players example, the template at helloworld.cc/ AIdrone2023).
(ai.io), which uses millions of data points snacksheet), because 2D arrays and lists It’s essential that our students understand
that AI is not a mystery, but comprises
computer programs just like any others,
RECOMMENDATION SYSTEMS ARE SEEN
“
which continue to follow rules that humans
program. Only then will the next generation
AS BLACK BOXES BECAUSE OF THE LARGE be confident enough to know that they have
NUMBER OF FACTORS THAT GO INTO THEM influence, and do not have to accept that AI
systems are beyond their control.

to select potential players for professional can be challenging for younger students.
clubs. Data such as athletic, technical, and The chance to modify a spreadsheet is
cognitive abilities are included, as well as much more appealing than inputting your
other relevant information such as a player’s chosen factors straight into a code editor,
age, height, weight, and playing history. especially when students have the power to
Another great example of this is YouTube. change the weighting of factors in a chart to
Many people have noticed that the longer something they feel is more realistic.
they use YouTube as a signed-in member,
the better their recommendations are. This Potential dangers
is because whenever you are signed in, In our snack recommendation system,
YouTube continues to collect data about the ranking multipliers of the factors are
what you clicked, how long you watched for, currently fixed. However, in many machine
and what other users who watched similar learning systems, these can be adjusted JAMES ABELA
videos to you also watched. Netflix recently by the program as data comes in. Even in James is head of computing at Garden
shared information on its key metrics, and our simple system of snacks, this carries International School in Kuala Lumpur,
these include: ‘starters’ (households that potential dangers, and these are important Malaysia. He is a Raspberry Pi Certified
watch two minutes of a film, or one episode points to discuss with students. The snacks Educator, founder of South East Asian
of a show), ‘completers’ (households that seen so far are not particularly healthy, and Computer Science Teachers Association,
watch 90 percent of a film or series season), if machine learning continues to run based and author of The Gamified Classroom.
and ‘watchers’ (households that watch purely on popularity, it might not suggest

helloworld.cc 47
 FEATURE

TIKTOK: A PATHWAY
FOR AI INCLUSION?
Drawing parallels from platforms like TikTok, we can envisage
a path for AI and data to address disparities

M
any ethical considerations This has real-world implications. For instance, as @mrspotatoqueen, uses TikTok to
surround the use of AI tools. It if a job recommendation algorithm is trained share her queer Muslim identity. Each of
is important to focus on how we can predominantly on data from male-dominated these content creators has amassed over
harness these tools to empower previously tech forums, it might inadvertently favour 100,000 followers.
disadvantaged groups. male candidates. Once ingrained in AI
systems, such biases can be deeply Incorporating diverse data
Bias in, bias out problematic, entrenching gender stereotypes Young people have embraced TikTok as
One example of bias in machine learning is further and perpetuating disparities. a place to create and consume content
the reliance on data sets that reflect historical that reflects their world. The majority of
contexts dominated by male perspectives. Al-powered communities TikTok’s active users are young people, with
This results in AI models that mirror these TikTok provides an example of a more 25 percent of them aged 10–19 and 57
biases. For instance, if a machine learning diverse form of data: user-generated percent identifying as female (helloworld.
model is trained primarily on male-centric content. The platform uses AI in two ways. cc/TikTokstats23). With accessible and
tech discussions, there’s a substantial risk Firstly, AI’s algorithms create data sets, affordable tools, TikTok encourages diverse
of it generating more male-centric tech similar to other recommender systems, content creation. A viral TikTok video,
answers. It’s a self-perpetuating cycle, where to recommend personalised content to when inclusive, resonates with a broader
biases in data lead to biases in tech solutions, consumers. Secondly, AI helps content audience, thus broadening the content
further perpetuating gender stereotypes. creators by suggesting enhancements users see. This has a domino effect, with the
(music, hashtags, etc.) to increase the media paying more attention to TikTok and
chances of their videos going viral. reflecting the wants and needs of the TikTok
What is unique about TikTok is that community in society.
the audience prefers diverse content and There are plenty of downsides to TikTok
seeks out videos that are relevant to them. too, such as pranks going too far and
TikTok’s 2023 trend report (helloworld. face filters promoting unrealistic beauty
cc/TikTokreport) states that: “TikTok’s standards. But TikTok is an example of AI
communities are relatable and supportive, powering inclusion. If we can feed AI more
which inspires people to make changes in diverse data, the resulting output will be
their lives.” more inclusive.
Disability educator India Atkinson Just as TikTok has managed to break
from Belfast, Northern Ireland, rose to down barriers, making content creation
TESSY MOSINDI TikTok fame after sharing videos about accessible and engaging to the masses, AI
Tessy, a learning manager at the Raspberry symbrachydactyly, a rare condition that tools have the potential to highlight,
Pi Foundation, is passionate about tackling caused her to be born without fingers on challenge, and even correct inherent biases.
the gender gap in computer science. her left hand; Garrison Hayes, a US-based Furthermore, tech platforms and tools
Previously a computer science teacher, she diversity, equity, and inclusion educator, used should showcase their value in improving
enjoys travelling and doing charity work. the platform to start a series on forgotten society, leveraging the evident enthusiasm
Black history in America; Shaz, also known for purpose-driven tech.

48 helloworld.cc
The free online platform
for teachers and students
around the world.

Free computer science resources, written

by specialist teachers

Save time when planning lessons and

marking work

Interactive Python coding questions

Real code examples in Python, C#, VB,

and Java

Tailored to GCSE and A level computer

science exam specifications

adacomputerscience.org
 FEATURE

DO THE BENEFITS OF AI
OUTWEIGH THE RISKS?
The genie is out of the bottle, but do the benefits of AI technology outweigh the risks?
We hear three different viewpoints from staff at the Raspberry Pi Foundation

T
aking time to reflect as an benefits of artificial intelligence (AI) arguments. Some of the team were
organisation and share differing technology outweigh the risks. swayed to be more cautious, while others
points of view is important at the Representatives were assigned to came to be more accepting of AI systems.
Raspberry Pi Foundation. We asked promote a neutral, positive, or negative Will the following arguments change your
everyone to consider whether the stance, then presented their best point of view?

© kubko/stock.adobe.com

50 helloworld.cc
 © nazarkru/stock.adobe.com
n From satnav to video games to shopping, AI technology is most likely already a part of your life

NEUTRAL BEN GARSIDE, EXPERIENCE AI PROGRAMME LEAD

We’re at an exciting place with AI, which still feels and as a society, we are generating and capturing
like a very new technological invention. But while data that can be used to train the next generation
that’s partly true, AI technology has in fact been of AI models.
in development for a long time. We can go all the This brings huge opportunities. AI systems are
way back to Alan Turing, who wrote a paper in able to process data, and recognise patterns in that
1950 called Computing Machinery and Intelligence data, far more quickly than humans can. I’m not
(helloworld.cc/turing1950). Despite the term ‘AI’ talking about AI systems that can spot the difference
being the media’s favourite buzzword at the moment, between cats and dogs, or play a game of chess. I’m
the reality is that we’ve all been living in an AI world thinking more about AI that advances science and
for quite a number of years. Do you have a smart benefits humanity. Google DeepMind, for example,
speaker at home? Have you used a satnav? What has found a way to use AI to predict a protein’s 3D
about a filter on social media? Then it’s very likely structure. In the near future, we might be curing
that you’ve interacted with an application that uses previously incurable diseases. We might be able
modern-day AI. to remove carbon from the Earth’s atmosphere research. It is rare that I ever agree with Elon, but I
The technologies I’ve just outlined aren’t efficiently. That’s something to be excited about. think he has a point. It’s unrealistic to expect anyone
worrying technological developments that we’ve Could this technology be dangerous in the to halt development, but time and effort should be
been scared of. Instead, they’ve crept up on us and wrong hands? Sure, but then so is nuclear fission. spent on reflecting on where we’re heading with this
become part of our normal. This is known as the AI Personally, what I am concerned about is finding technology, deciding what we want out of it, and
effect — once something becomes normal to us, we ourselves yet again in the hands of tech giants and making policies and laws that govern that.
stop thinking of it as AI. The digital calculator, for government policy makers. The speed of development We need collaborative efforts from policymakers,
example, was once considered artificial intelligence. of these systems is particularly rapid at the moment, researchers, and industry experts. We must
What’s driven the latest explosion of AI development and we’re in danger of entering a technological prioritise ethical considerations, and ensure that
is the place that we’ve naturally arrived at through arms race. ChatGPT hit the market in January, and we have robust regulations outlining transparency
other technological developments. We have more within weeks, Google and Microsoft were launching and accountability. If we foster that culture of
powerful computers than ever before; we have huge their own versions. People such as Elon Musk are responsible innovation, we can definitely harness the
data centres able to process vast quantities of data; saying that we should pause the development of AI benefits of AI while also mitigating the risks.

helloworld.cc 51
 FEATURE

YES MARK CALLEJA, LEARNING MANAGER

Humans have always been change-averse. Even when about job displacement and the erosion of
the benefits of a shift in thinking or new technology are traditional industries, as well as concerns about
obvious, we tend to treat it with suspicion and focus on technology exacerbating societal inequalities and
the negative aspects. This may be a prudent and cautious rendering education useless, with information now
course of action, but when consideration becomes panic, available to everyone everywhere. Sound familiar?
we tend to lose sight of the bigger picture.
Technopanics are not new, and stem mostly from a Technopanics really just force us to address our
lack of understanding, mixed with fear of the unknown. concerns about society through the frame of a new
Let me take you on a journey through time to illuminate technology and how it will exacerbate existing issues.
my point: Some say our ignorance of how AI works should create
concern … but we still don’t know how gravity works. We
n In 460 BCE, Plato worried that the invention of know what it does, and can reliably predict its operation,
the written word would ruin humans’ ability to but nobody knows how it does that. Ask a physicist and
remember things. Luckily, someone wrote this, and they’ll mumble something about bosons or loops or
all his other wisdom, down. strings, but that lack of understanding never stopped us
building houses, or planes, or a space station. from autonomous vehicles making our roads safer to
n In 1775, with the advent of the printed newspaper, Anthropomorphisation of AI (giving something AI-powered systems detecting fraud and cyberthreats.
it was thought that not having to go down to humanlike characteristics where they don’t exist) Ultimately, AI has the capacity to augment human
church and read the news off the door would tear makes it easy to demonise it. It’s not AI coming capabilities, opening doors to unimaginable possibilities
society apart, as everyone would now sit home for your job, it’s your boss trying to save money by and creating a brighter future for all of humanity.
alone to read the news. leveraging the power of this new technology. AI isn’t The deadliest thing humans have ever created,
going to increase the number of scams on the internet; which kills someone in the world every 24 seconds,
n In the early 1800s, the Luddites smashed weaving the criminals using it are. This framing tells us that the is the car, and most of us take the risk of using one
machines in fear of losing their jobs to them. outcome of technological change is inevitable and all without even thinking about it, because we know there
This didn’t stop progress, and Luddites are still a we can do is get governments to mop up the aftermath are rules in place for its use, and because we know
byword for technopanic today. The textile industry, of this ‘alien incursion’. the operator has been trained and legally assessed
meanwhile, now creates an amazing range of But how AI changes the world is far from as competent. The manufacturers are required to
incredible fabrics, from Gore-Tex to Kevlar to predetermined. It will depend on the balance of power pass stringent safety laws, and the vehicle must be
medical sutures. in millions of different workplaces; on legislation and maintained to a certain level. There are penalties for
regulatory enforcement; on the outcome of fights over their misuse. Anyone can take a car and ram it through
n At around the same time, rail travel was considered ideas, over laws, over working conditions, and over a crowd of people, or you can use it to take a bunch of
to be deadly; either you would suffocate at such the distribution of productivity gains, which are mind- children to the zoo. It’s not the vehicle that’s the risk;
insane speeds as 40mph, or women’s uteruses boggling in scope. it’s the humans.
would be sucked from their bodies. We now know AI has the potential to enhance productivity, We can never know all the answers, or foresee
that neither is true. automate tedious tasks, and optimise global resource exactly what might happen in the future. We can only
allocation. AI can unlock new frontiers in healthcare, do the best we can with what we have, to improve the
n In the 1990s, personal computers sparked a aiding in accurate diagnosis and personalised situation of everyone who lives and is yet to live on this
technopanic fuelled by fears of a destabilised treatment. It can supercharge education and create wet rock hurtling through space, mitigate the risks we
society in the face of rapid technological change. bespoke learning experiences targeted directly at the can, and do better when we know more — just as we
The rapid growth of the internet led to anxieties learner’s needs. AI can bolster safety and security, always have.

52 helloworld.cc
 © Sarawut/stock.adobe.com
n AI applications could potentially work without any human intervention to make real-world changes

NO JANE WAITE, SENIOR RESEARCH SCIENTIST

Right now, we, as a society, are not ready for AI. At this AI system ‘hallucinations’ (when large language
moment in time, the risks simply outweigh the benefits. models [LLMs] output information that does not make
Let’s start with job losses. A recent report estimated sense with respect to the training data), or there
that 400 to 800 million people would lose their jobs may be purposeful misinformation by governments,
to AI technologies by 2030, and in the US and parts organisations, or individual ‘bad actors’. Using the
of Europe, a third of people are expected to have to media for persuasion is not new, but are we aware
learn new skills to find different occupations. This is of how pervasive this is, and how we are being
a risk to all of us, working in a variety of industries, influenced? Should all systems be required to inform
from process-heavy work to creative and more highly us if AI technology is being used, and what the
skilled work. Over time, labour markets of course risks are? Maybe, but what international laws and
adjust to technological disruption, but in the short governance will enforce this, and how?
term, displaced workers need to find new work to avoid For me, though, the scariest risk is that of AI
rising unemployment rates and the downward spiral applications working with no human intervention to
these create. Following the coronavirus pandemic, and make real-world changes — for example, an AI system
observing the current state of the world economy, are sending emails, posting on social media, purchasing
we really well placed to support this? items, calling any other app, and even writing and and companies that could be used to build a weapon,
Secondly, the influence of the outputs of AI systems executing new programs. Can we even figure out including companies more likely to violate US export
is also a risk (Geoffrey Hinton, the grandfather of where things might go wrong? I’m not exaggerating restrictions.” (helloworld.cc/OpenAIreport)
neural networks, went as far as calling these an when I say that AI models have been used to suggest We are ill-prepared for AI. We lack a safety-critical
“existential threat”). Our connections with technology how best to commit crimes. No, this isn’t just sci- perspective at the developer, legislative, and societal
nudge our buying habits, our behaviours, our beliefs, fi; in Section 2.6 of OpenAI’s tests on GTP-4, they levels. AI models make things too easy for everyone:
and even our fundamental values. The outputs of state that LLMs can be used to “develop, acquire, or naive developers, organisations, bad actors, and
AI applications may be biased due to bias in their disperse … weapons” and that, “The model generated countries focused on land grabbing the power and
training data; they may be entirely false, through useful information about facility rentals, equipment, benefits of AI for themselves.

helloworld.cc 53
 OPINION

EXPLORING THE FUTURE:

PROGRAMMING WITH CHATGPT
Ken Kahn reports on his experiences creating apps by conversing with ChatGPT.
How will chatbots affect the learning of programming by schoolchildren?

S
ince the birth of the programming language need to install any software. Some of the apps I generated
Logo (helloworld.cc/Logo), computer were capable of performing the following tasks:
programming has played a fundamental role in
facilitating the acquisition of powerful ideas (helloworld. 1. Speaking random numbers with random language,
cc/Papert). Papert considered programming to be an pitch, rate, and voice, with subsequent enhancements
especially fertile environment for children, where they to incorporate user input (helloworld.cc/GPTnumbers)
might acquire powerful ideas such as problem 2. Drawing on a video feed in response to user gestures
decomposition, representation, abstraction, reflection, and spoken commands (helloworld.cc/GPTvideo)
and acquire the very concept of ideas being powerful. 3. Creating and training a neural network to predict
Clearly, a large overlap exists with the ideas of confidence from text data (helloworld.cc/GPTpredict)
computational thinking. 4. Simulating a conversation between two
Learning the concepts and details of programming is a personas, allowing user interjections
big task, but the constructionist community and proponents (helloworld.cc/GPTconversation)
of computational thinking consider it more than worthwhile. 5. Creating a water balloon game (helloworld.cc/
Their reasoning is that while creating programs from GPTgame)
nothing and debugging them can be hard, it’s a great 6. Simulating fireworks (helloworld.cc/GPTfireworks)
opportunity to learn some important skills. But now there is
an alternative: creating and debugging programs using an I never presented ChatGPT with a complete description
AI chatbot such as ChatGPT. of the desired final app. Instead, I started by
I wondered what this would mean for asking for a simplified version of what I
empowering learners as well as helping wanted. Once that was working, I asked
them to acquire powerful ideas. Here, I for enhancements one at a time. The
describe my experience of using ChatGPT experience was a bit like being the
to create sophisticated programs without ‘navigator’ giving guidance to the ‘driver’
writing a single line of a program. in pair programming (helloworld.cc/
pairprogramming). Perhaps ChatGPT
Programming with ChatGPT could manage a more complex task,
ChatGPT can generate programs in but more likely it would fail or produce
many programming languages. Since something different. Moreover,
ChatGPT is text-based, it is not as helpful incremental construction is more likely to
Ken Kahn recently retired from the
at generating Scratch or Snap! programs. lead to the user understanding how the
University of Oxford after a long
I used ChatGPT to generate JavaScript program works. For example, the ‘speak
career as a computer scientist
web applications, with most of the apps interested in children and computing. a random number’ app began as a button
involving AI capabilities. An advantage of He is now trying to understand how that would speak a random number. It
JavaScript is that JavaScript apps can be AI changes ideas about programming was then enhanced to display the number
run in any modern browser without the systems for children. and speech parameters, then to include a

54 helloworld.cc
way for the app to repeat what the user said with random

© ImageFlow/stock.adobe.com
speech settings, then to translate what was spoken to a
random language, and finally to add various user interface
and aesthetic improvements.
ChatGPT often generated code with bugs. I then
prompted the chatbot for a correction. Sometimes this had
to be repeated a few times before the bugs were fixed.
Sometimes the generated program failed completely with
no error message. Asking ChatGPT to enhance the program
so it would report errors reliably resulted in a version that
displayed error messages. Copying those error messages
into ChatGPT was sufficient to generate fixes to the
program’s problems.

Explanations
When ChatGPT generates code, it also adds explanatory
text. And if one doesn’t understand a generated function, n Children can use AI chatbots to design apps and creatively combine AI capabilities

one can request ChatGPT to add additional comments.

Typically, code is only lightly commented, but ChatGPT my experiments, a plug-in for ChatGPT named WebDev
produces very informative comments when these are was released. It can automate the process of copying and
requested. The resulting well-commented, neatly formatted pasting generated JavaScript and HTML into program files.
programs, displayed in five colours, are available for learners Computer programming has empowered children to
to copy, paste, read, or even question. One can also ask for creatively express themselves in powerful ways. Using an
explanations or a pseudocode version of the whole program. AI chatbot, students may no longer be creating programs
Importantly, GPT-4, the successor to GPT-3, can be prompted line by line or block by block, but they can still design apps
to generate a description of how the program works that is and even creatively combine AI capabilities. With fewer
meant to be understandable to a child of a specified age. technical skills, they can now construct more powerful
apps that rely upon AI capabilities, such as pose detection,
Is it actually programming? speech synthesis and recognition, vector encoding of texts,
There are myriad reasons why children should learn to neural network training, and much more.
program. I wonder how many of them still hold in an era As AI chatbots continue to be developed further, they will
where there is easy access to ChatGPT and the like. You undoubtedly become even more useful for students in
may ask whether a child who uses a chatbot to generate a creating increasingly sophisticated apps. While it is too early
program is still actually doing programming. I would argue to draw definitive conclusions about whether this approach
that the child is still programming — only they are doing so is the way forward for learning programming and acquiring
in English (or another human language a chatbot is trained powerful ideas, the questions raised warrant further
to generate). They are programming at a significantly investigation. If you’d like to read more about my
higher level of abstraction compared to using traditional conversations with ChatGPT, I’ve written a paper about my
programming languages. In this process, children are experience (helloworld.cc/Kahn).
expressing intentions instead of procedural instructions. It
is more like a dialogue where the child tries to state what
they want and also needs to provide the chatbot with useful
feedback about failures and errors.
THE CHATBOT DETAILS
For these experiments, I used the March 2023 version of
Empowering children ChatGPT Plus, based on the large language model GPT-4
Chatbots like GPT-4 arguably empower a larger number and costing $20/month. ChatGPT, the free version based
of children in a greater variety of tasks. For example, on GPT-3.5, performed less well in our tests, as did other
while I typed the prompts for the chatbot, they could be AI chatbots, such as Bard, Bing Chat, and Claude.
entered by speech using voice-typing features. After I did

helloworld.cc 55
 OPINION

AI: KEEP CALM AND CARRY ON

Richard Pawson implores computer science educators to not
lose sight of giving children programming opportunities

I
n 1981, Personal Computer World magazine Advocates for this new technology argue that this is
announced ‘the end of programming’, with an just the start for large language model (LLM) tools, which
exclusive story about the imminent launch of The are learning all the time, both with human intervention
Last One — a product that would supposedly translate and without it. Soon, they will be better than any human
your requirements into working code. The story was programmer, just like automated chess playing (something
picked up by the mainstream press and media. (I was the that did eventually become true, after having been
editor of a smaller, rival microcomputer magazine, which repeatedly forecast as imminent for nearly 50 years). In the
took a rather more sceptical view, repeatedly referring to context of education, some people are starting to say that
the putative product as ‘The Fast One’, which earned me we should de-emphasise the teaching of programming and
my first serious threat of a libel suit.) encourage children to learn how to use the AI-based code
The promise that computers would soon be doing all generators instead. I will now explore why this is a very
the programming for you is almost as old as automated dangerous idea, at several levels.
computing itself. The first attempt at what we would
now call a high-level programming language, in the early A step backwards
1950s, was grandly entitled Autocode (a name now Relegating the role of programming sounds remarkably
being reused by one of the new AI tools). And the COBOL like the idea that took root within British education in the
language evolved from Grace Hopper’s goal of mid-1990s: that the important thing was to teach
enabling people to define the functionality children how to use computers, not how to
of their business systems in plain English design them or program them. That decision
(helloworld.cc/GraceHopper). was eventually reversed about ten years
This is not to deny that the state of ago, by which time we had lost most of
the art has advanced gradually over the the generation of teachers that knew how
decades, with a dramatic acceleration to teach programming.
over the last couple of years. But I retain In a similar vein, Rishi Sunak, the
my scepticism. I know many senior current British prime minister, has
academic researchers who are working announced that he wants the UK to
with new AI coding tools and achieving become a leader in AI. But designing
impressive results, but it is important to and refining AI tools is strictly for the
Richard Pawson’s career began in
understand that they are experienced best-of-the-best in programming, and
1977, working for Commodore.
programmers who can recognise mistakes where are they going to come from? I’m
After 40 years in the computing
in generated code and guide the AI tool afraid I do not subscribe to the view that
industry, he became a computer
accordingly, in an iterative process. Like science teacher at Stowe School in programming is for everyone. But if we
most power tools — both software and Buckingham, UK. He now writes free don’t give our children the opportunity
physical — they make the skilled user A-level teaching resources, and is a to find out whether they are or aren’t
better, and the unskilled user obvious. regular contributor to CAS. well suited to programming, we are

56 helloworld.cc
 © Gorodenkoff/stock.adobe.com
n Learning to program is the most compelling aspect of computing as an academic subject

certainly going to end up with a very much smaller pool of your program finally runs correctly, is the compelling and
programming talent. unique feature of computing as an academic subject. Let’s
Let’s suppose that the enthusiasts for LLM tools are not lose sight of that.
right. If so, the threat of AI isn’t just to programmers, but to Going back to my opening story, the announcement of
virtually all knowledge workers, including writers, lawyers, The Last One took place just before the first main wave of
most business jobs, and, of course, teachers. We might the introduction of computing into schools with the
as well abandon education altogether. I am old enough
to remember when IT started to seriously threaten jobs,
and the UK government’s response was that everyone
POWER TOOLS MAKE THE
should go to university so that we could become a nation
of knowledge workers. The reality today is that more of our
“ SKILLED USER BETTER AND
graduates end up as baristas than barristers.
The teaching of programming has faced multiple threats
THE UNSKILLED USER OBVIOUS
in recent years. It is threatened by the idea that what’s
most important is computational thinking (a fancy name
for problem-solving, which has always been a key part BBC Micro, to which so many professional programmers
of multiple better-established school subjects, including in the UK subsequently owed their careers. Just think what
maths, sciences, design technology, and art), and by its would have happened if we had allowed the threat of the
partner in crime ‘computing unplugged’. The unique thing imminent demise of the programmer to take root back then.
about real programming is that you can design a solution So, my plea, in the face of LLM tools, is this: whether you
to a problem once, in executable form, and then you can see them primarily as a threat or an opportunity, keep calm
solve different instances of the same problem multiple and carry on … teaching programming. And make it your
times, even millions of times, for zero extra effort. goal to improve your own programming skills, and strive
Learning to program, and the reward that comes when continuously for better ways to teach them.

helloworld.cc 57
 FEATURE

STARTING A CODE CLUB:

STORIES FROM NEW CLUBS
Code Club met with four clubs in the UK that shared their experiences of starting
their own Code Club: Beverley School — a specialist school for autistic students;
Joseph Turner Primary — a mainstream school; St Paul’s C of E Academy — also
a mainstream school; and Fab Lab — a maker space in a community library

C
ode Club is a global network of Q: What inspired you to start a Code Club own ideas of what they wanted to make
free after-school coding clubs led in your school? and began to choose their own projects.
by teachers and/or volunteers for learners A: I was motivated by attending a Code Some use Raspberry Pi Foundation projects
aged 9 to 13 years. It’s easy to start a Club event. I met lots of amazing people and some do their own thing. We started
coding club and you don’t need to be who showcased all the support for starting printing out projects as this was easier for
a coding expert. To get a better idea of and running a club. Also, the children at our the students to work with. Then we adapted
what it’s like to start a club, check out the school have a passion for computing, and by them using Widgit Symbols (helloworld.
following four clubs and their experiences. running a club we have helped to give them cc/widgets) to make them accessible to
an environment to grow their skills. everyone in the club.
Code Club at Beverley School
Beverley School is a special community Q: What was the process of setting up Q: What have you learnt that might be
school for autistic children in Middlesbrough, the club like? helpful to share with a new Code Club?
UK. Bethany Gibson, the club leader, A: It was simple — we registered on the A: Our biggest challenge when we started
attended a Code Club workshop in January Code Club website and then got started was supporting our students in creating
2023 and went on to start her Code Club with sessions. accounts for Scratch. We now spend the
soon after. Bethany feels that running a Code first week with new coders getting them
Club has allowed everyone’s confidence to Q: What projects have you been using? signed up to everything we offer. Plus, we
grow. She explains how brilliant it’s been to A: We started off following the project play some fun unplugged computer-themed
see her students supporting each other, both paths (helloworld.cc/projectpaths), but games or activities. This way, the following
in the club and in other areas of school life. the students quickly started to have their weeks run much smoother for the students.
I’m enjoying seeing club members
develop their independence. Some of our
students have created whole games that
they work on in the club and at home. This
is something we didn’t really see until we
started the club.

Code Club at Joseph Turner Primary

At Joseph Turner Primary School in Tipton,
UK, Scott Sefton is the club leader and
also the school’s computing lead. In the
spring term of 2023, he began his club
with twelve Year 6 learners (aged 10–11)
and then moved on to twelve Year 5 (aged
9–10) learners in the summer term. Since
both groups were new to coding, Scott
found it helpful to demonstrate how to code
the projects while encouraging students to
n Elena and her young coders enjoying a Code Club session at St Paul’s Church of England Academy share ideas and help each other.

58 helloworld.cc
Q: Why did you decide to launch a Code university when I studied computer science. Q: What advice would you give a new Code
Club in your school? I wanted to give children a chance to share Club based on your experience?
A: My head teacher tasked us with setting between year groups and almost become A: Our biggest hurdle while we were starting
up this club as part of a whole-school ambassadors in coding, to be able to share out was planning and getting a structure in
expansion into more clubs. I have continued skills they’ve learnt in Code Club with their place for the flow of the sessions. Now we
it because of demand for the club. I hope peers in class. start with a ‘free’ time period where the kids
to continue developing programming can complete the prior week’s projects or
understanding throughout Joseph Turner Q: Do you have any advice that could try to make their own, and we have a show-
Primary School. benefit a new Code Club? and-tell time at the end to give the kids time
A: Plan out a schedule for the term so you to express themselves and share what they
Q: How have you found the Code are prepped. Spend time practising how to have been making.
Club projects? type and log in at the start of each session
A: The resources are excellent for us to because a lot of children struggle with the Q: What are you currently working on?
use. I find the projects good; they are well- basic skills of using a computer and that A: Currently we are working through the
detailed, easy to follow, and the children may hinder their work later. Have a mixed Scratch pathways. While you don’t normally
range of ages and mix the children up in need to dedicate so much time to one
your Code Club so they can help each language, we thought it would be better
I WANT TO INSPIRE
“ ALL CHILDREN TO
other out.

Q: What impact are you hoping for as your

for the kids to learn all they can from one
[language] at the same pace together
before moving on to more complex and
BE INTERESTED Code Club continues to grow? challenging languages.

IN COMPUTING A: I would hope that when our children

leave for secondary school, they go into
We have found the show and tell to
be very successful. It boosts our learners’
their computing lessons with a passion confidence, giving them the opportunity to
really enjoy them, especially when they pick for computing. I also want coding to be show their parents and ourselves what they
what they want to do for that week. something children feel they excel in, like have learnt. It also inspires new ideas and
English and Maths, as so many jobs rely on ways of coding in other young people.
Q: What have you learnt that you could programming these days.
share with a new Code Club? Divine, aged eight, had this to say about Interested in Code Club?
A: I live model the projects so children their Code Club at St Paul’s: If you’d like to learn more about setting up a
can see how to code accurately. By live “I enjoy that you get to do different Code Club, please register your interest on
modelling, I mean I read the instructions activities and it’s really nice. And the teacher this form: rpf.io/interest. You never know,
aloud with the children, and then show them is really nice and kind.” you could be the next coding club leader in
what to do on Scratch itself. your community!
Stella, aged ten, had this to say about Code Club at Fab Lab, Hartlepool
their Code Club at Joseph Turner: The Code Club at Fab Lab, based in the
“I love Code Club because we can do Hartlepool Community Hubs, UK, has been
different animations and stuff.” running since early 2021. Michael Storey,
one of the volunteers at the club, told us
Code Club at St Paul’s C of E Academy about his experience.
St Paul’s Church of England Academy,
also in Tipton, launched its Code Club in Q: When you began your journey with
May 2023. Elena Lockett, the club leader Code Club, how did you find the projects?
and computing lead, has twelve learners A: While getting our footing, the premade
attending her club, the majority being girls. projects and plans on the Code Club website
She values a child-led and peer learning were a huge help when deciding what we
environment and therefore has a mixed wanted to cover and teach that week. They
group of club members, ranging from Year 3 also helped the kids to work at their own
RUJEKO MOYO
to Year 6 (aged 7–11). pace, as we needed to be prepared for new
Rujeko Moyo works at the Raspberry
arrivals every week. The difficulty selector
Pi Foundation as the Code Club
Q: What motivated you to start a Code (i.e. Level 1, 2, 3) on the overall projects
Community Coordinator for England.
Club in your school? page helped us pick random projects that
She works with schools and libraries
A: I wanted to inspire all children to be were fun and would use our 3D printers
across England, supporting them to
interested in computing — especially girls or other resources we have, like the LEGO
start and run their own Code Clubs.
— as I wrote my dissertation on this at Education WeDo kits.

helloworld.cc 59
 FEATURE

CREATIVE COORDINATE
GEOMETRY AND CT
Gitika Kaur shares how computing and maths practitioners can collaborate to
integrate STEM lessons using computational thinking as a connecting link

A
s a former software engineer and components: computational thinking (CT), I will share the curriculum design and the
a long-time teacher of maths and Scratch programming, and coordinate incredible work of the students.
computing, I have always believed there geometry. We implemented the curriculum
is a strong relationship between coding online, with a group of students aged eight Curriculum design
in computer science and problem-solving to twelve, for a period of five weeks in the Our curriculum design is underpinned
in maths. As part of my master’s in STEM spring term. The first four weeks included by Seymour Papert’s constructionist
education, I had the opportunity to model five hours of teaching and learning. The framework, which emphasises ‘learning
these thoughts. I undertook educational fifth week concluded the curriculum with by making’ (helloworld.cc/papert1980).
design research with a maths teacher an hour-long assessment through the The activities within the curriculum were
from India, with the aim of creating a presentation of students’ projects and a curated to guide students towards creating
curriculum that would bring together three short formative assessment. In this article, their own ‘virtual artefacts’ at the end of the
course — for example, unique and creative
n Table 1 Computer science
and maths concepts geometrical shapes including repetitive
can be linked through
computational thinking
patterns of triangles. While mapping Scratch
programming concepts with coordinate
geometry concepts, I used computational
thinking as a problem-solving tool to link
them together. Table 1 gives a flavour of the
relationship between the three components
and the expected learning outcomes.
The pedagogical approach focused on
transforming the role of a teacher from an
instructor to a constructionist facilitator, who:

• Articulates the clear purpose of each facet

of computational thinking (decomposition,
pattern recognition, abstraction,
and algorithm)
• Provides a balanced amount of scaffolding
— both stepping in and diving back — so
that learners neither lose the joy to learn
nor feel incompetent and get distracted
(helloworld.cc/resnick2013)
• Induces activities that encourage
student collaboration
• Devises ‘low-floor and high-ceiling’
activities, and uses phrases like, ‘Let’s
construct a solution together’ to prompt
students to construct their own piece of
code for solving problems, rather than
seeking one finite solution

60 helloworld.cc
n Post-intervention results showed a marked improvement over pre-intervention results

Results of the intervention use computational thinking facets for solving

The findings indicated that as students problems.” However, she also mentioned
explored the movement of a sprite on their that she could foresee some challenges
own, they engaged in using programming related to in-service training for learning to
concepts to find the changing coordinate. program, because she felt less confident
While exploring the concept of negative and facilitating students to debug their code.
positive axes, they engaged in answering the Though I cannot generalise the teachers’
exam-based questions correctly, themselves, perspectives, as I could interview only
n An eight-year-old student exhibited creativity when determining the
without the need for those concepts to be one maths teacher, I cannot neglect two relationship between the length of the diagonal and the sides of a square

taught. As a computer teacher, I focused important facts: firstly, the ten years
on facilitating the improvement of students’ of teaching experience of the teacher
understanding of block coding techniques, participant and secondly, the finding in the
in order to empower them to achieve their literature that there exists a “knowledge
desired outcomes. Through purposeful and skill gap, especially with respect to
teaching sessions and assessment through subject-based pedagogy” (helloworld.
their project presentations and a short cc/Curzon). Therefore, I suggest that in
formative test, it became evident that order to efficiently implement CT-infused
student exploration and engagement lessons in the maths classroom, we need
significantly enhanced their response time to deploy computer teachers along with
and problem-solving skills. maths teachers. This was borne out by the
It is worth noting that, by the end of our maths teacher I interviewed, who went on to
intervention, all students demonstrated say that, “The collaboration with computer
their proficiency in defining computational teachers is mandatory to teach computer GITIKA KAUR
thinking and displayed confidence in utilising science concepts, otherwise the skill gap of Gitika is a teacher of computer science at
its facets when confronted with assessment teachers will limit the students’ calibres also.” Desborough College in Maidenhead, UK. She
questions. As educators, we were deeply has spent the last seven years working as an
moved by the students’ outstanding Takeaway for teachers educator and consultant in India, teaching
performances throughout this process. I suggest that teachers need to take care to and acknowledging the scope of computing in
We take pride in the fact that our role “combine diving in with stepping back” creating an exploratory learning environment.
shifted from being instructors to facilitators, (helloworld.cc/resnick2013), to maintain an
allowing the students to take ownership equilibrium between teaching and facilitating
of their learning journey and thrive in their before steering learners towards taking concepts and link them together using
academic endeavours. ownership of their learning. More research is computational thinking. Future rigorous
The participating maths teacher added, needed into the implementation of the quality testing of the intervention, through
“I am amazed to see the confidence and computational thinking curriculum and the partnerships between researchers and
agility of students to play with sprite[s] and creation of CT-infused lessons through teachers, is needed in order to develop,
explore all the mathematical concepts with effective collaboration between computer validate, and assess the impact of the
interest and ease. I look forward to using science teachers and maths teachers, to curated lesson plans on learners’ creativity
Scratch in my classroom such that students create lessons that map disciplinary and problem-solving skills.

helloworld.cc 61
 FEATURE

ONLINE COURSES FOR

PROFESSIONAL GROWTH
Tips for unlocking your potential through continuing professional development

O
nline courses are a fantastic
option for teachers to engage in
continuing professional development
(CPD) in a way that can fit in with even
the busiest of lifestyles. There are a
huge number of quality online learning
opportunities, which you can complete in
your own time and at your own pace.
Online courses in particular offer
collaboration tools for learners to work
together. Although the communication
is not face-to-face, discussions provide
opportunities to engage with a diverse
group of fellow educators around the world.
In this article, we will outline a range of
approaches to help you make the most of
online courses, including those provided by
the Raspberry Pi Foundation, which will be
useful whether you are considering taking n Combining text and images in
your first online course or are a veteran of notes can embed learning

the format.
the time and wherever possible stick to it. Take notes
Choose your online course You can usually complete online courses at Even while learning online, non-digital note-
Before you begin a course, make sure it is your own pace — our edX courses can be taking is really important. We recommend
the right one for you. Most online courses, completed by committing as little as one that you take notes throughout online
including those on edX (helloworld.cc/ hour per week. If you can, find a quiet place courses to reinforce your learning and as
edXRPF), include a short description of to learn from where you won’t be distracted a study aid when it’s time to revise prior
the course content. Make sure you choose so that you can focus on your learning. to taking any assessments. Plus, you can
a path that will meet your expectations always go back to your notes long after you
and help you develop your skills as an Use videos have completed a course.
educator. Look out for courses that are Online courses commonly include videos Don’t be afraid to get creative with your
appropriate for the age range of your featuring a presenter. These videos can note-taking, whether it’s through sketching
learners, or that address specific areas of often include animations to help illustrate or highlighting key information. For
teaching and learning that you would like to a concept or a screencast showing the inspiration, pictured above is a set of sketch
improve upon. presenter working through a problem. In notes from our Programming 101: An
all of these cases, you should feel free to Introduction to Python for Educators course
Craft your ideal learning schedule pause the video and rewind it to help you (helloworld.cc/Programming101).
This is completely personal to you, but it’s understand a particular point, or to follow
a good idea to establish when and where along with an example. For accessibility Build a learning community
you can best dedicate your time to online reasons, courses should include all of the Many online courses include opportunities
learning. This may be a regular time, every key content from videos in text form, often for you to share your thoughts and
week in the same place, or you may need as a transcript. You can use this text instead progress. Although it may sound scary
to be more flexible to fit in with your other of the video, or as additional support during to put your ideas and work onto a public
commitments. Whatever you choose, plan or after watching the video. forum, sharing and discussing with your

62 helloworld.cc
 n Ben Hall presenting an online course

fellow learners can really help to enhance to stop and think about what you’ve learnt. online courses on subjects ranging from
your learning journey. Remember that You can link what you’ve learnt to what you programming in Python and Scratch, web
you aren’t alone — your fellow learners already know, or consider different contexts development and design, cybersecurity,
are in the same position, and a friendly to which your new knowledge could apply. to machine learning and AI on the
comment can go a long way. By posting, You could also think about how you have edX platform.
and commenting on other people’s posts, been learning, and if there are any changes We hope this article helps you make the
you help to build a sense of community, you want to make to your approach to most of online learning for CPD, and that
meaning you and your fellow learners feel your learning. you enjoy learning with us.
more involved and supported. You might also find it helpful to create
your own summary of what you’ve learnt
Ask for help when you complete a course.
There will likely be people on your course
from a variety of backgrounds who will be Sign up now!
able to offer support and advice — make The Raspberry Pi Foundation has a long
the most of this! If you are struggling history of providing world-class online CPD,
to grasp a particular concept, there will and we are delighted to announce that
probably be someone else who can help we’ve joined edX, a global online learning
you; don’t be afraid to ask. The more platform. Through our free online courses,
specific you can be with your questions, the we enable any educator to learn the skills
more likely others may be able to help. You required to teach students about computing MICHAEL CONTERIO
may also be able to support other learners and how to create with digital technologies. AND BEN HALL
yourself, which is a great way to put your Since 2017, over 250,000 people have Michael is a former physicist and now
learning into practice. taken our online courses, including 19,000 works as an online course production
teachers in England alone. The move manager at the Raspberry Pi Foundation.
Reflect on what you’ve learnt to edX builds on this success to help us Ben is a learning manager at the Raspberry
It’s important to consider what you’ve learnt bring high-quality training to many more Pi Foundation. He is a CAS Master Teacher
to help embed your learning. Good courses teachers worldwide. and a Raspberry Pi Certified Educator
will give you specific chances to reflect, but You can now sign up (helloworld. (@hengehall).
even if these aren’t there, try to find places cc/edXRPF) for 20 of our most popular

helloworld.cc 63
 FEATURE

TEACH FIRST:
TRAINEE COMPUTING TEACHERS
Daljit Shoker and Rachel Arthur invite first-year trainee teachers to share
how they put computing pedagogies into practice at Teach First

I
n Hello World issue 21 (helloworld. When Rachel wrote the computing delivering the computing curriculum, Daljit
cc/issue21), Daljit Shoker and curriculum for Teach First, her goal was takes a look at the transformative journey
Rachel Arthur shared how they deliver to create a curriculum that was practical, undertaken by three first-year trainee
computing pedagogies to their first-year engaging, and accessible. The Teach First computer science teachers. Katy Barber,
trainee computer science teachers. Now, computing curriculum includes twelve Lindsay Gibbons, and Cara Griffiths have
this article looks at how Teach First trainees pedagogical principles (helloworld. embraced the PRIMM pedagogy (Predict–
have implemented those pedagogies in cc/12pps), equipping trainee teachers with Run–Investigate–Modify–Make) in their
their teaching practice and the key research the tools they need to make a lasting impact classrooms, leading to remarkable outcomes
they used in their literature reviews in their on students who experience educational with engagement, code comprehension,
computing assignments. disadvantage. Building on the success of handwriting code, and debugging code.

64 helloworld.cc
KATY BARBER
As part of my assignment, I explored several key understood BIDMAS (brackets, indices, division, The final assessment revealed a potential PRIMM
pieces of pedagogical research, including PRIMM, multiplication, addition, and subtraction), a new code pitfall. Written exam-style questions diverged from the
Parson’s Problems, pair programming, and live coding. block posed challenges, particularly for those with lower PRIMM format, unsettling students accustomed to IDEs.
I decided to focus on one of my classes (aged 14–16) reading skills or English as an additional language (EAL). Many struggled with syntax, highlighting the need to
who were struggling with understanding Python syntax; Here, I found that scaffolding was vital for their progress. adapt PRIMM for traditional exams. A key learning for
this was acting as a barrier to them understanding how The second episode featured a monthly exam-style me here was incorporating early handwritten coding
to program efficiently. Over the past few months, assessment with help sheets to aid with writing correct practice to better equip students.
I have been exploring how to deploy PRIMM to syntax. The biggest challenge I identified here was Overall, I found that my reflections offered valuable
structure my programming lessons and help improve with students struggling with error identification due insights into the application of PRIMM within the
my teaching and consequent pupil understanding of to integrated development environment (IDE) programming curriculum. While PRIMM proves
Python syntax. Throughout my assignment, I focused dependency — this assessment was paper-based. effective in fostering comprehension, adaptation is
on three main teaching episodes and reflected on how The third reflection revolved around a lesson on crucial for aligning with the broader exam-based goals
I had delivered the components of PRIMM and the subroutines, featuring teacher-led exposition and of students. Balancing IDE usage with handwritten
effect it had had. worksheets and promoting independent engagement code practice emerged as a good strategy to prepare
The first lesson focused on arithmetic expressions, during the MAKE stage. Here, I noticed that tailored students holistically for both classroom learning and
using PRIMM with teacher-led prediction and scaffolding led to notable improvement for students future assessments, and has impacted my future
independent investigation. While most students who had previously struggled. planning of how to deliver programming lessons.

LINDSAY GIBBONS
For trainee teachers on the Teach First training Big Book of Computing Pedagogy (helloworld.cc/ that they could see how the code could be written and
programme, the first assignment was to pick one BBpedagogy) — with the aim of making coding more debugged, and especially how the logic of deciding what
topic in the computer science curriculum and dive in accessible to this mixed-ability class. Using PRIMM path to take was formulated.
deep to the research and good existing practice. In my gave consistency to the students, allowing them to pick Once more code writing was taking place in
specific case, the focus was on programming lessons their level and engage more in the lessons. the lessons, I realised the importance of ‘directed
for students aged 14–16 years. For many students, PRIMM provided a structure for the slow student talking’ in the classroom, so I introduced pair
computer programming is among the most challenging introduction of key topics, which gave the students programming. These collaboration sessions allowed the
aspects of computer science. From the start of confidence with trying new syntax and coding students to talk about how the code was being used.
teaching Python to the class, it became clear that techniques, and encouraged them to engage. Reflecting Finally, during the second part of the teaching year,
there was a reluctance to engage with programming. on the lessons after a term of using PRIMM, I realised I saw a huge leap in terms of engagement. I introduced
This cohort of students had little real-life experience there was a group of students who were stagnating at a Use–Modify–Create examples from The Little Book of
of programming to put it into context; the ideas behind specific point in the lessons — they didn’t want to push Algorithms 2.0 (helloworld.cc/LBOA2) so that the
coding were very abstract. From my literature research themselves past the easy PREDICT stage. I introduced students got more experience of editing and running
and after discussions with my subject development several other pedagogies alongside PRIMM. I found Python code. They could get real debugging experience
lead (SDL) at Teach First, I chose PRIMM — from The that I needed to be more visible in creating code, so and start taking ownership of their code.

helloworld.cc 65
 FEATURE

CARA GRIFFITHS
Through the opportunities provided to me by Teach is something that is often overlooked in academic
First, I was able to take some of my theoretical papers — students are not just learning a new
learning and put it into practice in the classroom concept; they are discovering a new way of learning,
last year. Parson’s Problems are activities that are and must be able to practise it before they can
often overlooked by teachers, as they are usually successfully learn with it. Doing this as a tabletop
tagged on to the end of a lesson when time needs to activity also helped greatly, as students connected
DALJIT SHOKER
Daljit is a subject development lead for
be used up. Instead, I wanted to look at how turning with the physicality of being able to move paper
them into a fully unplugged activity could best help into the correct positions.
computing at Teach First. She delivers
learners aged 11–13 who were just beginning their Based on all of this, I would like to share my
the computing curriculum to trainee
journey into learning Python. top tips for successfully using Parson’s Problems computer science teachers in the UK and
The use of Parson’s Problems proved incredibly in your lessons: supports them with developing their subject
effective in helping students understand code — knowledge and teaching approaches.
but particularly when they were used in a specific n Make sure that the activity is done on paper first Previously, Daljit taught ICT/computing for
way. What I found was that students who were just n Always start with a scenario that is familiar to 18 years (@DaljitShoker).
given jumbled-up lines of code to sort found the the students
task confusing and too difficult. However, when n When using code, start small — four lines of
we started with a Parson’s Problem example that code can be very effective
focused on an everyday scenario (brushing teeth,
for example), the students found it much easier I hope that by sharing this research, I can help
to grasp the activity and were more successful others to feel more comfortable using Parson’s
with the code. What we realised from doing this Problems in their lessons. Happy solving!

THESE STRATEGIES ARE ABOUT INSPIRING

“ A LIFELONG PASSION FOR CODING AND
TECHNOLOGY IN THE MINDS OF STUDENTS RACHEL ARTHUR
Rachel is the head of computing, programme
design, at Teach First. She has previously
We hope you enjoyed reading about this deeper understanding of coding concepts, been a head of computing and assistant
educational journey and glimpsed the encourage problem-solving skills, and head teacher in schools across the UK
power of putting into practice the create an inclusive and supportive (@rarthurtweets).
computing pedagogies shared by these classroom environment where all students
three teachers during their first year of have the opportunity to thrive. So, as you
teaching. If you are currently teaching or embark on your teaching journey,
planning to teach programming, why not remember that these strategies are not just Katy Barber, Lindsay Gibbons,
embrace these innovative approaches? about imparting knowledge; they are about and Cara Griffiths are newly qualified
They can undoubtedly make the process of inspiring a lifelong passion for coding and computer science teachers who have
learning programming more engaging and technology in the minds of your students. successfully completed the first year of the
effective. By integrating PRIMM, Parson’s Happy teaching, and may your classroom Teach First training programme and are now
Problems, and unplugged activities into be filled with the excitement of discovery in their second year of teaching.
your teaching repertoire, you can foster a and the joy of learning!

66 helloworld.cc
 FEATURE

n With support from Google, four teachers attended the WiPSCE ‘23 conference

WIPSCE 2023: A COMMUNITY OF

RESEARCHERS AND TEACHERS
Insights from four teachers attending the WiPSCE conference for the first time

T
he annual WiPSCE Conference In September 2023, the University of
on Primary and Secondary Cambridge Department of Computer
Computing Education Research is an Science and Technology and the
international conference where university Raspberry Pi Computing Education
academics present their research on the Research Centre hosted WiPSCE 2023.
teaching and learning of computing in As part of this, we were delighted to
primary and secondary schools. A range welcome Google as a sponsor of the
of peer-reviewed papers and posters are event. Google has a deep commitment to
presented, and all are then published in computing education, especially within
the Association for Computing Machinery communities that have been historically
(ACM) digital library (helloworld.cc/ underserved. Through partnerships and
WiPSCE23). Topics covered include programs, Google supports organisations,
curriculum, pedagogy, assessment, universities, and schools globally to ensure
equity, tools, and teacher education. that every student has the opportunity
There is a long history of WiPSCE to access the benefits of a computing
welcoming teachers to be part of the education to help shape their future. JANE WAITE
conference. Teachers are invited to hear With support from Google, WiPSCE Jane is the senior research scientist at
about research and to discuss it with the offered free places for five UK computing the Raspberry Pi Computing Education
researchers and their fellow teachers, teachers to attend the conference. Teachers Research Centre. She was previously a
providing an exciting and collaborative were selected from primary and secondary schoolteacher and an IT developer, and now
environment for sharing ideas and state schools from five different areas she investigates the teaching and learning
initiatives that are highly relevant of England. Meet four of the teacher’s of computing (@janewaite).
to classroom practice. selected to attend WiPSCE 2023.

helloworld.cc 67
FEATURE

ANDY COLLEY LAURUS CHEADLE HULME, CHEADLE

What did you study and where? How would you describe the conference?
ICT at Salford. Educational! It really opened my eyes to the way that
academic research and institutions function. It was a
How did you end up in computer science? really friendly environment; I felt safe to ask questions
I’m a converted ICT teacher. and have my thinking challenged.

Why were you interested in attending WiPSCE? Whose research were you inspired or
I have a background as an Advanced Skills Teacher surprised by?
(AST) and Lead Practitioner for Teaching and Learning Hint content for block coding (helloworld.cc/
(T&L). I’ve always been interested in best practice Greifenstein) — I wasn’t aware of LitterBox
pedagogy and research-informed practice. It seemed (helloworld.cc/litterbox). Anette Benz made me
like a great chance to meet the people doing the aware of different lenses for identifying student n The research that seemed the most focused/useful
studies and gain a better understanding of how preferences (helloworld.cc/Bentz). was where researchers collaborated directly
research happens. with teachers
What were your key takeaways? n The human story behind why we use technology
Have you been to a conference like this before? n It’s useful for teachers to have an appreciation of can help engage learners with the more
No, I haven’t. how and why research is happening abstract concepts

RACHAEL COULTART
ST NICHOLAS C OF E PRIMARY SCHOOL AND NURSERY, STEVENAGE

What did you study and where? Whose research were you inspired or surprised by?
I’ve done a BA in Education at Durham University and an n Michael Rücker: A framework to consider the
MA in Education at the University of Hertfordshire. impact of technology sounds so sensible for
considering how to help our learners understand
How did you end up in computer science? the consequences of what they are doing
No one else wanted to do it, and my husband was ‘good (helloworld.cc/Rucker).
with computers’! n Luisa Greifenstein: Loved the online seminar
about LitterBox — dabbled with it straight after
Why were you interested in attending WiPSCE? (helloworld.cc/Greifenstein).
Having been involved in three different action research n Anette Bentz: Both presentations were fascinating.
projects over the last few years and it being the best Thinking about patterners and dramatists as
continuing professional development I’ve ever had, I’ve opposed to girls and boys (helloworld.cc/Bentz).
become increasingly interested in how research can Looking at ways of developing understanding of
influence my practice. loops also made me want to look at my practice in
more detail (helloworld.cc/Bentz-loops).
Have you been to a conference like this before? n Anaclara Gerosa: Definitely want to explore What were your key takeaways?
Never! I’ve attended conferences where research is 'grounded cognition' theory some more. But would Research is a messy business, delivering more
referenced and applied to classroom practice, but have also like to see research that explores whole-body questions than answers! It’s OK to see text-based
never heard about research projects direct from the analysis (helloworld.cc/Gerosa). coding like Python as ASCII art and I shouldn’t worry/
researchers, which puts a very different spin on it. n Katharine Childs: I’d love to do something like concern myself about transitioning KS2 pupils to
the 'Hour of Code' following a teacher exchange it. Teacher advocacy and building a community
How would you describe the conference? I did in The Gambia many years ago to lean on are key in developing and delivering an
Much more academic (and a different pace) than (helloworld.cc/Childs). equitable curriculum. Understanding and talking about
I’m used to. A fantastic place to reflect on one’s n Isabella Gransbury: I have used the driver- differing human values is perhaps the key to helping
own practice and evaluate what you’re doing in navigator approach a lot, but love the sound of our learners understand the impact of technology
the classroom. driver-driver too (helloworld.cc/Gransbury). on society.

68 helloworld.cc
IAN GRAY
ENDSLEIGH HOLY CHILD VC ACADEMY, HULL

What did you study and where? How would you describe the conference?
BA in English Literature & Educational Studies; PGCE in Very informative. Meeting people from different
Primary — both at Hull University. education systems around the world has been inspiring
and a welcome opportunity to leave the English
How did you end up in computer science? education system bubble.
I became a computing lead because I had an interest in
technology and coding, and nobody else wanted to do it. Whose research were you inspired by?
Michael Kölling, Joanna Goode, Michael Rücker,
Why were you interested in attending WiPSCE? Anette Bentz, Karen Nolan, Anja Marx, Kate
I’ve always been interested in research and how the Farrell, and Veronica Catete (helloworld.cc/
‘why’ relates to the ‘how’ in learning. I wanted to WiPSCEproceedings).
gain an understanding of how research is presented,
worked on, and conducted, as I would like to conduct What were your key takeaways?
my own research. n Education systems around the world are working n Agile is useful in helping students to collaborate
towards similar goals but in different ways n Use of driver–driver programming approach
Have you been to a conference like this before? n Using a structural approach to teaching (collaborating on code)
No, only the SEND in Computing conference at the York programming could help develop better coders n Pupils' interests and how they can affect their
Computing Hub. n Design research approaches can work in approaches (patterner/balanced/dramatist)
the classroom n Designing for equity

SOPHIE FENN PATE’S GRAMMAR SCHOOL, CHELTENHAM

What did you study and where? pose questions. Despite my limited familiarity with
I studied Computer Science at the University of research terminology and statistical terms, I realised
Warwick to master's level. that our insights as educators held equal value in these
discussions. This process demystified the research
Why were you interested in attending WiPSCE? process, so I’m now initiating a small classroom-based
I discovered WiPSCE through the ACM membership and research project myself!
was excited to learn about a conference dedicated to
primary and secondary computing education research. Whose research were you inspired or surprised by?
I wanted to find out what research was being carried I found Michael Kölling's presentation on a frame-based
out and hoped to get some new ideas to enhance my editor for Python fascinating (helloworld.cc/Kolling).
teaching practices. This innovative approach allows programmers to focus
on program structure and fundamental concepts rather
Have you been to a conference like this before? than syntax. I found it incredibly thought-provoking
I’ve attended edTech conferences in the past and that something we deal with every day without thinking
similar commercial events, but entering a research- twice can be reimagined. It encourages me to consider
focused conference like WiPSCE was a completely new how we might approach teaching challenges differently. seldom straightforward. Context plays a crucial role,
experience for me. from students’ prior exposure to computing to differing
What were your key takeaways? interpretations of what computer science entails
How would you describe the conference? Research is not solely about large-scale projects among students, teachers, and researchers. Therefore,
I would describe the conference as an incredibly involving thousands of students. A substantial portion of testing research findings in your classroom with your
friendly and inclusive environment. Following each the research involves just a few students and students is essential to evaluate their applicability in
presentation, there is time to discuss the research and investigating small changes to practice. Findings are your unique context.

Note: We will be writing articles about each of the referenced research items and will update the links as we publish.

helloworld.cc 69
 LESSON PLAN

INFORMATION
TECHNOLOGY
The following lesson
plan is taken from
the ‘Computing systems and

AROUND US
networks — IT around us’ unit
of The Computing Curriculum
(TCC), written by the Raspberry
Pi Foundation. It is aimed at
learners aged six to seven, and
is designed to promote learners’ Help learners explore IT in familiar environments
understanding of the various
aspects of using a computer to
T
his is the third lesson in The Computing
create and change text. AGE RANGE Curriculum’s ‘Computing systems and
networks — IT around us’ unit for learners aged
6-7 years six to seven. During this lesson, learners will
identify information technology (IT) that they

ABOUT THE interact with in familiar settings, such as home,

COMPUTING
shops, and streets. They will talk about the uses
OBJECTIVES of IT in these environments and understand how
CURRICULUM IT is used in many workplaces.
Find examples of IT

Sort IT by where it is found ACTIVITY 1: WHERE ARE

YOU LIKELY TO FIND IT?
Talk about uses of IT 5 MINUTES

REQUIREMENTS
Ask learners to think about where IT is likely to
be found. Remind learners that IT is anything
The Computing Curriculum is
that is a computer or works with a computer. Ask
the Raspberry Pi Foundation’s
‘A2 Activity sheet – IT learns to show a thumbs up if they think that IT
bank of free lesson plans and
other resources that offer in public places’ from is likely to be found in the place described, and
thumbs down if it’s unlikely that IT would be
educators everything they need helloworld.cc/ITintheworld found there. They can show their thumbs in the
to teach learners aged 5–16.
middle (sideways) if they are unsure.
It covers the full breadth of
computing, including computing ‘A3 Activity sheet – Talk Is IT likely to be found in any of these places?

systems, programming, creating about IT’ from helloworld. n A shop

media, data and information,
cc/ITintheworld n The woods
and societal impacts of
n Work offices
digital technology.
n The beach
Every unit of work contains a
unit overview; a learning graph After learners respond to each description, have
to show the progression of a brief discussion about their thoughts.
skills and concepts in a unit; Note: IT could potentially be found in all of
and lesson content, including a these locations, e.g. a person on the beach with
lesson plan, slides, and formative a phone. The discussion should be open and
assessment opportunities. Find consider where IT is more likely to be found
them when you sign up for a free and why.
account at helloworld.cc/tcc.

70 helloworld.cc
n Activity 2 What are these IT objects?

ACTIVITY 2: IT IN PUBLIC PLACES 20 MINUTES

Give learners the ‘A2 Activity sheet — IT in learners if they know what the object is.
public places’ or display the activity sheet on If learners have some knowledge of an
the interactive board for the class. object, ask them to explain its purpose.
Discuss the two locations — a busy street Next, ask the learners to sort which IT is
and a café or restaurant — pictured on likely to be found in each location. Tell them
the activity sheet where IT could be used. to reflect on the previous discussions and try
Encourage the learners to think–pair–share to place the IT object in the location where
some of the IT that they might find in either it is most likely to be found. If learners are
of these places. unsure about an object, they can discuss it
Look at each picture of IT on the activity with another learner, or choose not to place it
sheet or the interactive board and ask in either location.

PLENARY
ACTIVITY:
IT EVERYWHERE
3 MINUTES

Remind learners that IT

is used in almost every
n Activity 3 Talk about IT workplace. Ask learners if they
can think of a job in which IT
ACTIVITY 3: TALK ABOUT IT 10 MINUTES may not be used. Most jobs
will involve some level of IT,
Provide learners with ‘A3 Activity sheet — 1. Where IT is found even if it’s only a phone.
Talk about IT’ or display the activity sheet on 2. What it is used for
the interactive board for the class. Highlight 3. How it works

RELEVANT LINKS
that the devices shown — traffic lights and
a pedestrian crossing — are found in lots Scaffolding opportunity: instead of focusing
of places. Ask the learners to think–pair–share on a pedestrian crossing, learners could
about the devices shown. discuss the sequence of lights used for a set TCC ‘IT in the world’ lesson:
They should describe: of traffic lights. helloworld.cc/ITintheworld

helloworld.cc 71
 LESSON PLAN

AGE RANGE
11–14 years WHAT IS AI?
Introduce your students to data-driven artificial intelligence
OBJECTIVES and how it is used in the world around them

D
 escribe the difference
T
his lesson is taken from the Raspberry and how it is being used in the world around

between ‘data-driven’ and Pi Foundation’s Experience AI them. Students will also consider some of the
resources (experience-ai.org). The aim is to advantages and disadvantages of
‘rule-based’ approaches to help students explore the current state of (AI) AI systems.
application development

N
 ame examples of AI
STARTER: WHAT IS INTELLIGENCE? 3 MINS
applications
As students enter the classroom, display the instead, reveal the definition in the slide deck:

O
 utline some benefits question ‘What is intelligence?’ on your ‘Commonly, people describe intelligence
board. Students should discuss their as the ability to learn and adapt/react to
and issues of using AI thoughts with the person next to them. This new situations.’ Highlight to students that the
applications part of the lesson aims to set the scene key point is that intelligence is the capability
and to focus students’ attention. You do not to learn something and use that knowledge
need to take answers from the students; to react and adapt to new situations.
REQUIREMENTS
R
 esources from lesson
INTRODUCTION:
one of the Experience AI
resources (helloworld.
THE ‘INTELLIGENT’ PIECE OF PAPER 10 MINS

cc/lesson1) ‘The intelligent piece of paper’ is an Encourage the students to share their
adapted version of an activity created by thoughts. Someone might point out that it
Paul Curzon and Peter McOwan of Queen is not the piece of paper that is intelligent,
n Slides Mary University of London with support but rather the person who wrote the
from the Engineering and Physical Sciences instructions. The person who wrote the
n ‘The intelligent piece Research Council (EPSRC) and Google algorithm is intelligent because they have
(helloworld.cc/intelligentpaper). learned about IF/THEN rules and can
of paper’ resource
To play this game, you will need a apply them to create an algorithm that
(printed copy for volunteer who will challenge the paper in cannot lose in tic-tac-toe. They can also

demonstration) play a game of tic-tac-toe (noughts and apply those IF/THEN rules to solve other
crosses). Another volunteer will act as the problems.
paper and follow the instructions given by To further the discussion, ask the class,
n Lesson one worksheet ‘The intelligent piece of paper’ worksheet. “If a computer follows an algorithm written
The game will result in either a draw or a by a human, does that make the computer
c raiyon.com allowlisted on win for the intelligent piece of paper. After intelligent?” The answer should be no. The
the game ends, ask the class, “Does that computer is not learning or reacting to new
computers
make the piece of paper intelligent?” situations; it is simply following IF/THEN rules.

72 helloworld.cc
n Figure 1

ACTIVITY 1: WHAT IS ARTIFICIAL INTELLIGENCE? 12 MINS

The first task for students in this activity is the traditional approach to programming,
to use their worksheet to record what they where an algorithm is followed to solve a
think is meant by artificial intelligence. This computational problem. In contrast, a data-
worksheet will be used throughout the driven approach is a way of building artificial
lesson, so keep it handy! Play a video to the intelligence systems using statistics from vast
class where the experts at Google DeepMind quantities of data, rather than writing out the
explain what is meant by AI and address rules in a program.
some common misconceptions. Using the example of creating an AI model
Next, describe the rule-based and data- to play a game of chess, ask students to
driven approaches to creating applications consider why a data-driven approach would
(Figure 1). The rule-based approach is be more beneficial.

ACTIVITY 2: GENERATIVE AI 10 MINS

In this activity students will get hands-on some examples). Whilst students are
and use an online application to generate experimenting, circulate around the class and
their own artwork. Before they get creative, prompt students to consider the following:
outline the meaning of generative AI
(applications designed to generate ‘creative’ n Try generating the image again with the
content such as sound, images, and text). same prompts and the same art style.
Then direct students to the worksheet Does it produce the same artwork as
where they will find a link to craiyon.com last time?
to create their artwork. They must create n Try changing your prompts. Does it work
artwork for either: better with fewer or more words?
n Who might find this technology useful?
n A poster for the world climate n How might artists feel about this
change conference technology? Do you think they might
n The wall of a café or restaurant be worried?
n Who owns the artwork you create? You or
Allow students five minutes to use the the creators of the application?
application (images can take around a n What are the potential issues with using
minute to generate — see Figure 2 for AI to generate art?

helloworld.cc 73
 LESSON PLAN

n Figure 2

ACTIVITY 3: COMPUTER VISION 10 MINS

Using slide 18 listed in the resources, explain n Why do you think there is a confidence
to students that computer vision is a field rating? Is that important?
of AI that attempts to extract meaningful n Why do you think the application is more
information from images. After that, ask the confident about some elements of images
students to refer to their worksheets and than others?
explore a computer vision online application n Who might find this technology useful?
and answer the questions on the worksheet n How important is the confidence rating for
(helloworld.cc/computer-vision). Once on a driverless car, for example?
the website, ask students to select one of the n In what ways might visually impaired
sample images to see what the AI system people benefit from this technology?
can identify in it. Allow four minutes for n Can you see this technology
students to experiment. being misused?
Whilst students are experimenting, n Would you feel comfortable knowing
circulate around the class and use the that you can be personally identified by
following questions to prompt discussion a camera when you are walking around,
about what they are looking at: for example?

PLENARY ACTIVITY: AI OR NOT AI? 10 MINS

RELEVANT LINKS
The purpose of this task is for students to or not it uses AI. Highlight that it may not be
reflect on what they have learnt in this lesson clear to them, in which case they can answer
helloworld.cc/intelligentpaper
and to apply their knowledge to see if they ‘Could be AI’. Explain that they must attempt
craiyon.com
can identify when AI might be being used. to justify their answer for each application.
helloworld.cc/computer-vision Ask them to evaluate each application Remind students to think about whether
experience-ai.org listed on their worksheet and decide whether each example is rule-based or data-driven.

74 helloworld.cc
 BEBRAS PUZZLE

THE BEBRAS ABOUT BEBRAS

PUZZLE PAGE
Bebras is organised in over 55 countries and aims to get
students excited about computing and computational
thinking. Last November, over 365,000 students
participated in the UK annual challenge.
Our archived questions let you create
Students who do well in UK Bebras get invited to take part in your own automarking quizzes at any time
during the year. To find out more and to
the Oxford University Computing Challenge (OUCC); this issue’s
register your school, head to bebras.uk.
Bebras problem is taken from the first round of the 2023 OUCC

THE PROBLEM: PIG LATIN DOMAIN COMPUTING

Looping and selection
Algorithms and
programming KEYWORD
SKILLS SPOTLIGHT:
Algorithmic thinking
ARTIFICIAL
AGE
14–18 years
INTELLIGENCE
DIFFICULTY RATING Defining words and phrases
Ages 14–16 medium; in computer science
ages 16–18 easy
I asked ChatGPT to define
artificial intelligence. Here is the
Pig Latin is a language game made by Output format: a string consisting of the start of what it came up with:
modifying English words: translated input
Example input: Hello internet “AI, or artificial intelligence, refers
n If the word starts with a consonant (for Example output: ello-hay internet-yay to the creation and development
example ‘hello’): of intelligent machines and
Remove the first letter of the word This problem was originally written by software systems that can
(‘ello’) the OUCC team from Jamaica. Solution on simulate human intelligence.”
Add a “-” to the end of the word page 85.
(‘ello-‘) We need to be clear that AIs do not
Add the removed letter (‘ello-h’) Further information have self-awareness or understand
Add “ay” (‘ello-hay’) If you want to try programming this in the what they are doing like humans.
n If the word starts with a vowel (for interface used for OUCC, you can do so at They are simply following algorithms
example ‘internet’): helloworld.cc/piglatin. The OUCC system and mathematical constructs to
Add “-yay” to the end of the word supports eleven programming languages, create output that is similar to
(‘internet-yay’) including those most used in schools. The what we would produce. Because
students’ solutions are sent to our testing most humans do not understand
Using this information, write a program that server, where they are run using several the inner workings of AI, there is a
will translate a sentence into Pig Latin. different inputs to test that students have danger that we anthropomorphise
thought of all the boundary cases. Partial them, using language such
Constraints: the provided input word will consist scores are awarded and students can make as, “intelligent machines” and
of upper-case and lower-case letters only. several attempts. The tasks are then archived “language understanding”. They
Input format: an English sentence or group for schools with a UK Bebras account to use in are just very powerful tools
of words. their own automarking quizzes. humans have built to aid us.

helloworld.cc 75
 FEATURE

© ViDi Studio/stock.adobe.com
n Learning the skills to create movies also has a
positive effect in other areas of the curriculum

MAKING MOVIES USING GREEN-

SCREEN TECHNOLOGY
Gemma Coleman catches up with primary teacher Seema Zerafa about her experiences
developing learners’ movie-making skills with a film competition

hat is Childnet and the Childnet

W
learning and e-safety. The idea behind this is
Film Competition? that pupils who want to enter the competition
Childnet is a charity that works to make the are equipped with the skills and knowledge
internet a great and safe place for children to do so, and even those who don’t want to
and young people. For the last 14 years, the take part are learning to conceive and
charity has run an annual film competition, produce digital content that is effective,
open to all young people aged 7–18 in the compliant, and secure.
UK (helloworld.cc/childnetcomp). The brief I’ve carried out this competition with pupils
is to create a two-minute film or storyboard aged nine to eleven for the last three years.
that focuses on a set online safety theme, However, we’re effectively preparing
which changes each year. This year’s theme students from the age of six, exposing them
was ‘Time to talk — how can people to green-screen technology, which I often link
support each other online?’ to teaching around topic days such as World
SEEMA ZERAFA Book Day, so by the age of eight or so, they
Seema is the head of digital learning at How do you prepare for and run the are familiar with the technology.
Pembridge Hall School in London, UK competition? I split the project into an eight-week
(Instagram: The competition launches on Safer Internet module, one lesson per week. We divide the
@creativecomputing_missm). Day every February (saferinternet.org.uk), so class into groups of four at most and, during
I dedicate the spring term to project-based the first four weeks, we learn about each app

76 helloworld.cc
and how to use them effectively. The second You can make a green screen from green adding in a voice note, which is a super new
half of the block is the ‘workshop’, when the cloth or backing paper, which you may feature. It’s amazing how much enjoyment
groups create a storyboard, and record and already have in the art department. and excitement you can add to learning to
edit a movie. At the end, they can decide code when you’re combining it with
whether they want to enter the competition, What skills does the competition movie-making.
either by using what they’ve produced, or help build? Learning the skills to create movies also
starting from scratch. I’ve found the competition to be the perfect has a really positive domino effect in other
launch pad to embed movie-making skills areas of the curriculum. There is so much
What tools do you use for the and important messages about online safety potential for cross-curricular collaboration.
competition? into the curriculum. For example, after pupils have learnt how to
I recommend researching which apps you While acquiring movie-making skills, the use iMovie, could they make an iMovie about
feel would work best for the project. I use students develop a range of key digital a religion in humanities? Or could they use
Green Screen by Do Ink (helloworld.cc/ literacy skills during the project. It’s easy for a GarageBand to make music for a movie in
doink), which you can buy for a small fee, and student to get carried away with design ideas drama or music class?
iMovie (helloworld.cc/imovie), which you can and themes, but we introduce the students to
use on an iPad. They are both pretty the concept of a ‘target audience’ and seeing How do you assess students’ work for
foolproof; you don’t need to be an IT specialist if their choices fit that target audience. We a competition like this?
to use them, and both feature a mini tutorial also spend time looking at branding and I suggest setting the criteria at the start of
when you download them. Some Windows seeing whether their products align with the the project (video duration, theme, format,
10 alternatives include Clipchamp, which is a brand, and use similar fonts and colours. etc.) and compiling a checklist of
pre-installed app with any Windows 10 Although we celebrate a different theme considerations for students (copyright issues,
device (helloworld.cc/Clipchamp); or every year, we always address data music and image licensing, digital consent,
Microsoft Flip is a great web-based program protection, for example by obtaining online etc.). Throughout the process, I undertake
to use (helloworld.cc/flip). It allows students consent. We discuss who should appear in formative assessment; I don’t just assess the
to create mini videos safely, and it includes their movies, and the importance of asking filming, but I also appraise teamwork and
child-friendly filters and emojis. The best way permission, especially when featuring a communication.
to explain it is that it’s just like creating videos child’s face online. We also discuss whether
on TikTok, but monitored and controlled by a to display names on the website, who could What are your top tips for educators
teacher. Microsoft Flip includes tools for the see this information, and whether we would who want to try this competition out?
teacher to limit the duration of the video and want them to see it. Our discussions revolve Don’t underestimate your students! Being
switch off features, such as face filters and around determining what information is safe digital natives, they pick up skills really
emojis, if students don’t use them effectively. to post, and what isn’t. quickly and can produce content that is often
It’s also useful to use collaborative tools like You can embed all sorts of key computing better than that created by adults. They also
Microsoft Teams, Google Docs, and Microsoft skills into the project, including programming respond really well to being given
Whiteboard. We set up dedicated Microsoft skills. For example, we use Marty the Robot responsibility and are far more likely to take
Teams channels so pupils can communicate (a walking robot that is fully programmable the task seriously.
about the project with their peers. and can be connected over Wi-Fi or I’m passionate about the power of play, so
You don’t need to spend money on a green Bluetooth, robotical.io) as a character in our I’d recommend giving students time to play
screen; anything works as long as it’s green! I movies. To help Marty perform actions and with the apps you’ve chosen. It’s important
would also suggest looking at the resources move, pupils can construct their own code for them to experiment independently and
you already have at your disposal in school. using MartyBlocks and MartyBlocks Jr, even see what they can achieve. They often teach
me how to do things, even though I’m an IT
specialist — if someone asks me something I
don’t know the answer to, I open it out to the
class. They love teaching each other.
Finally, remember to fit in time to allow
students to conduct their own research, and
make sure you provide feedback on their
plans and films so they have enough time to
edit. I always ensure I give examples of what
both good and bad content looks like, and
discuss what characterises the two.
Helpful information and resources are free
to download in the Childnet entry packs
n Fully programmable robots are the lead characters in a green-screen film (helloworld.cc/childnetcomppack).

helloworld.cc 77
CONVERSATION INSIDER’S GUIDE

WORK-RELATED
LEARNING EXPERIENCE
In this Insider’s Guide, Alan O’Donohoe shares how educators and businesses
can work together to offer work-related learning opportunities for students

I
t’s essential that all young people are capable of making weeks, late in Year 10 or early Year 11 (ages 14–16). Without
informed decisions about their future, particularly those exploring all the reasons here, there’s been a trend for schools
who lack the necessary support at home. Many young people to gradually move away from this model. There was a necessary
will not have had a real experience of working until they are and obvious decline because of Covid-19 lockdowns, but there’s
18 years old at the earliest, yet they are making decisions no sign of a return to previous levels yet. The business I work for,
that can affect their future many years before then. Through
productive partnerships with employers, schools can do a lot
to support young people through a programme of work-related
learning experiences.
GOOD CAREER GUIDANCE
I’ve written this Hello World guide with an audience of
educators and employers in mind, with the aim of providing The Gatsby Benchmarks were published in 2013 and serve as a framework
low-cost, practical suggestions for employer-hosted events with for exceptional careers provision. The UK government’s Careers Strategy has
measurable impact that can be adapted for use. This guide will adopted these benchmarks (helloworld.cc/Gatsby) and they have proven
highlight the range of potential opportunities available. When to have a positive effect on students’ readiness and achievements. Although
approaching an employer, the suggestions in the article can act four of the benchmarks (below) are strongly linked to employer engagement,
as a catalyst to start meaningful conversations to achieve similar it probably makes more practical sense to focus on benchmarks 2 and 4 in
levels of success as we’ve had with our events. school with support from employers, and to focus on benchmarks 5 and 6
when visiting a workplace.
Traditional work experience
To be clear, this guide doesn’t really include the more traditional 2. Learning from career and labour market information: supports
work experience placements, which typically take place over two students to make informed decisions about their future and be prepared to
enter the workforce.

ALAN O’DONOHOE 4. Linking curriculum learning to careers: helps students see the real-life
With 30 years of experience teaching and applications of their education and inspires them to set career goals early.
leading in schools in northern England,
Alan leads The Exa Foundation (exa. 5. Encounters with employers and employees: can give students valuable
foundation) on a mission to inspire insights into various career paths and job expectations, and help them build
digital makers, support the teaching of professional networks, which can enhance their employability.
computing, and promote the appropriate
usage of technology (@exafoundation, 6. Experiences of workplaces: allows students to develop practical skills,
@teknoteacher). understand workplace culture, and make informed career choices.

78 helloworld.cc
 n Alan giving an Exa induction

Exa Networks based in Bradford, is an internet service provider

with about 50 employees. Over a recent summer holiday period,
we had one student on work experience placement. The student
FEEDBACK
(age 16) organised the placement, but that is an extremely rare
occurrence nowadays. n C
 hristine, Director of ICT & Digital Learning, Leeds City Academy: “Our
When our business witnessed a reduction in enquiries for Year 10 students (ages 14–15) loved meeting the employees and finding
work experience placements, we started to review what we do out more about their jobs and pathways. They enjoyed setting up the
to support young people in terms of work-related learning. Our Raspberry Pi computers, building in Minecraft, the pair programming, and
business is fairly unique in that we have offered school-based the awarding of raffle tickets for praise worked really well.”
workshops nationwide over the last decade. This is separate from n M
 ohammed Shazad Bhatti, Bradford College: “Thank you for a wonderful
our core business, which is providing internet connectivity and experience for our learners. The general feedback was that the students
online protection to schools. People are often surprised to learn enjoyed everything, especially the network security element.”
how a small tech business in Bradford, a city in northern England, n S
 tudent at Bradford College: “It was a really good session, as I have
is dedicated to providing an education and careers programme for learnt a lot of things. I have learnt in more detail how cyberattackers
schools, and this has been recognised through impact awards. attack victims. I have understood more deeply the four cornerstones of
computational thinking and have some clear examples of how these are
Half-day visits applied to real-life situations.”
When our normal event programme resumed following Covid, we n A
 nother student at Bradford College: “The session was fun because of the
switched to including work-related experience days for young hands-on activities. We learnt a new term called ‘cyberthreat actor’. In
people as a more significant part of our programme offering. general, the session was very engaging and moved at a very good pace. The
For the next academic year, we’ve committed to offering two most important part was that everyone was involved.”
opportunities every month for schools in Bradford to book a visit,

helloworld.cc 79
CONVERSATION INSIDER’S GUIDE

RESOURCES
STEM Ambassadors: SkillsHouse:
Teachers have the opportunity to enhance work-related learning opportunities Exa Networks has been working with SkillsHouse, a local team under Bradford
for their students by using the resources of STEM Ambassadors. With over Council’s Employment and Skills services. SkillsHouse aims to enrich learning
30,000 individuals eager to share their expertise and enthusiasm, this well- experiences by aligning with Gatsby Benchmarks and providing access to
established volunteer programme can offer valuable insights, career talks, and employer-led work-related learning activities through their SkillsHouse Portal.
engaging activities to inspire and educate young minds. STEM Ambassadors are Readers in other areas may be able to find similar agencies managed by their
trained and supported to deliver a wide array of activities that resonate with local council (skillshouse.co.uk). Andrew Gray at SkillsHouse Bradford, says:
students. Schools can easily request a STEM Ambassador through the online “EXA and similar organisations are deeply committed to their corporate social
platform, which can bring real-world relevance into classrooms and empower responsibility, put simply as ‘giving back and paying forward’. We’d encourage
students for success in STEM-related careers (helloworld.cc/ambassadors). others to become similarly engaged.”

n Students taking a break from the morning activities

80 helloworld.cc
and we are currently exploring opportunities to expand the work but yes — they earn enough to manage their way of life. When
we do nationwide. In the last few years, school groups typically we ask the young people if they’d like to work at Exa in the future,
visit our business for a half-day or longer to take part in a series of there’s always an overwhelmingly positive response. This activity
challenges and activities that help them make tangible connections is often the easiest to manage, but seems to have the most
between what they’ve learnt in school and the workplace. significant impact on the young people who attend. Initially, young
people are shy about asking the first questions, so I have some
Mini induction previously asked questions prepared, and then they soon want to
During the visits, participants are first shown around part of ask their own.
the building as a mini induction to fire and safety regulations,
something all of our staff receive on their first day. Our building Impact and feedback
has controlled access zones and young people need to Students and teachers have told us they found these events
understand where these are and why. This is followed by a fun, incredibly valuable, particularly as the young people attending
interactive game in which participants have to guess facts about (ages 8–18) have had an opportunity to see people like them and
the business. Later, in an information security session, participants whom they can relate to, working in fulfilling roles, and this has
are tested on their knowledge of security threats and steps to helped positively influence career and study choices.
prevent them. When we’ve hosted events for a longer period of time with the
same group, we have the opportunity to enhance to experience
Meet the employees and make it even more immersive. It gives us more time to
During the visit, colleagues from the business spend time with the incorporate a variety of work-related activities.
visitors explaining their roles within their team, followed by a Q&A The events have provided opportunities for young people to
session. Young people typically want to know how much they spend some time interacting with work colleagues and learning
earn and what grades they received at school. The students are more about the business in general. By hosting whole teaching
visibly impressed to hear employees relate how their experience groups for these events, we’re able to offer the experience to a
of community, recognition, and fulfilment trumps a high salary, larger number of young people than before.

SUGGESTED ACTIVITIES
ACTIVITY 1: ONE FALSE, TWO TRUE been hacked”, and is eliminated from the game. Students must deduce the hackers’
On page 85 of Hello World magazine issue 20 (helloworld.cc/20), there is an identities through reasoning, nominating suspects, and voting. The player with the
interactive activity designed to develop computational thinking skills. The activity most votes is removed, and their true identity disclosed. Repeat the process until all
encourages students to apply computational thinking to real-life scenarios by cyberthreat actors are identified and eliminated.
identifying the false statement among three presented facts. This captivating game immerses students in computational thinking while
The activity starts with an employer sharing three brief facts about themselves promoting teamwork and strategic thinking. Employers can use it to explain how
or the company, with only two being true. In small groups, students work together it relates to their security policies and real-world threats. For a more detailed
to identify the false fact by devising questions. Emphasise a computer science explanation, see the original article in issue 20.
perspective, prompting logical reasoning and abstraction in their approach. During
the Q&A session, students are encouraged to actively listen and use computational ACTIVITY 3: WHAT’S IN THE BAG?
thinking skills. At the end of the activity, the false statement is revealed, and This is an engaging and fun cybersecurity activity for students. A bag contains cards
feedback is provided on their question types. that have technical phrases printed on them, including cybersecurity-related terms
such as threats and protectors, and unrelated terms like cat breeds. The teacher selects
ACTIVITY 2: HACK ATTACK a card at random, and students must swiftly categorise it as a threat, a protector, or a
On page 86 of issue 20 (helloworld.cc/20), there is a description of an activity cat. Teams can earn points by correctly identifying threats or protectors, and earn three
that requires students to collaboratively identify and eliminate imaginary points for identifying a cat. To score extra points, teams can explain the meaning of the
cyberthreat actors using computational thinking skills. selected term and its relevance. For example, correctly identifying ‘phishing’ as a threat,
This activity, inspired by the game Werewolf, is easy to set up and does not explaining how it works, and ways to safeguard against it, can earn two points.
require any equipment. It is best played with a large group of players, and students As the game progresses, the teacher or employer can introduce more complex
use logical deduction and pattern recognition to succeed. The moderator, initially terms and deeper questions to enhance students’ understanding of cybersecurity.
the teacher, secretly designates three players as cyberthreat actors. These actors Non-cybersecurity terms add an amusing challenge to the game, and students can
silently choose victims for a coordinated cyberattack during the night phase. In the return them to the bag if drawn. This activity encourages critical thinking about
daytime phase, the victim is revealed, unable to communicate beyond saying, “I’ve cybersecurity with a touch of light-hearted fun.

helloworld.cc 81
CONVERSATION PERSONAL ACCOUNT

TEACHING T LEVELS
Gemma Coleman catches up with Halima Bhayat about her experience of
running T levels and looks at how you can get started in your school or college

hat are T levels and how do they work?

W
Once the tasks start, students are not allowed to use the
T levels are qualifications that students in England internet. We give them set code, data files, and test log templates.
generally take between the ages of 16 and 18. The government Our scheme of learning covers a mock period as well as an
introduced them in 2020 to give students another pathway into example taught step-by-step before this assessment. We
further education or the world of work. found that students appreciated having hands-on experience of
The course is structured over two academic years. In the first practising this section.
year, students take exams and an employer-set project, while in the
second year, students complete a 67-hour occupational-specialism And what is the occupational-specialism project?
project. Students also need to complete a 315-hour placement in This assessment takes up the full second year of the course.
industry over the two years. You need to really plan how you want to run this academic year,
including when you may want to hold mock exams. The project
Why did you choose to introduce them to your school? involves four tasks, with flexibility regarding when educators run
We were one of the first pilot schools to run a T-level course. We each task. Students can use the internet for the duration of this
wanted to take part because the course would open students up assessment and are therefore able to independently solve issues
to real-life work experiences in fields they’re interested in. We and create code using support from the internet.
also knew it would mean employers would be more able to offer The four tasks are split like a system development cycle, where
apprenticeships post-school because of students working with for the first task, students research and plan a model. Then in the
them, and we liked that this provided options other than university second task, they implement and create the model, before testing
for pupils. the model in task three. Finally, in task four they evaluate what they
We also liked that having a project in both years meant students have implemented and suggest further improvements.
were learning skills, not just theory for paper exams. At the time,
there were only three subjects on offer. We chose the Digital What does the placement in industry involve? And what
Production, Design and Development course as it offered three are the benefits?
strands (computing, digital skills, and business) with a broad You are required to organise a placement in industry covering a
category of options to teach. We already had staff with knowledge minimum of 315 hours, of which 20 percent can be remote. We
and skills built in those areas, and we structured our timetable for found this section of the course particularly challenging as a lot
staff to create most of the course around computing, followed by of companies, especially in the digital sector, are now working
digital skills, and business. entirely remotely. To make this a little easier, my tip would be to
build a network of people who work in industry, including parents,
Could you describe the employer-set project a little more? employers, and local organisations who might be able to support
The project is a task-based controlled assessment, split into five you in taking students on for a placement.
sections using the systems design life cycle. The style of this project We decided to cover our placement section in two-week blocks, as
work has allowed pupils to think beyond the classroom and work this was the most comfortable amount of time for both students and
using the same methods as real software developers. The full the employers. We set the placement times at the start and end of
assessment lasts 14.5 hours and takes place over specific days. the year, allowing plenty of time in between for staff and students to
We provide students with a scenario (for example, a coffee shop complete other schoolwork and to fit the placements around exams.
wants to become computerised and have a front-end and back-end So far, we have worked with IBM and small local businesses.
system for its orders) as a pretask, which they can discuss with During their placements, students end up doing all sorts of
their peers and spend time researching. things. For example, some developed a chatbot that focused on

82 helloworld.cc
 ©Emma Smedley
n T levels combine classroom-based learning
with experience in the real world of work

sustainability and the safe disposal of materials. To help them to work together to support the course, from careers team, finance
reflect on their experiences, we created diary logs where they could staff, and technicians through to the school senior leadership team.
record their new-found skills and knowledge and connect them I would also suggest keeping your technical team involved in
back to their school projects. the process of the course. The employer-set project, for example,
involves students downloading and using programming language
What have you found to be the biggest benefits of the libraries (such as Pandas), so you might need tech support!
course? And what have students enjoyed the most? Throughout the year, we’ve invested in new technology like
Much of the T-level course we offer involves independent work, robots, VR headsets, and drones to enhance our students’ learning
with students building their skills and problem-solving tasks experience. We’ve also organised employer workshops, visits,
themselves. We’ve found this helps students prepare themselves talks, and trips. The course is all about helping students to grow
for further education and the workforce. as individuals and develop their aspirations. We’ve found that this
Overall, our students have been positive about the course. Not holistic approach to education has really helped.
only have they gained valuable experience, but the placements
have also opened up doors to paid degree apprenticeships.
This has been especially helpful for our most vulnerable and HALIMA BHAYAT
underrepresented students, including those from ethnic minority Halima has over 20 years of
backgrounds, SEND students, and female students. Unfortunately, experience working in education,
there is still a shortage of women, especially from ethnic minority and is currently the head of
backgrounds, in these fields, which makes it difficult for employers computing and digital T levels
to find them. Thankfully, the course has helped these groups at Ursuline High School in
connect with potential employers, which will hopefully help close the UK. Halima is an Asian
the gap for them in the tech industry. Women of Achievement 2021
Finalist and the Computing at
What advice would you offer to schools interested in School (CAS) Merton lead for
pursuing T levels? all schools. She is a computing
Planning each of the long assessments has been a challenge, advocate, empowering ethnic
especially as other state examinations and mocks take place at minority girls to pursue a
the same time. It has been key for us to plan the exams as early as STEAM education and career.
possible in the autumn term to ensure we’re organised! Everyone has

helloworld.cc 83
CONVERSATION EVOLVING COMPUTING

PREPARING YOUNG PEOPLE

TO UNDERSTAND SYSTEMS
Michael Conterio and Tracy Gardner discuss how to prepare young people to interact
with the computing systems they will encounter in their lives

W
hen we first introduce the idea of computer systems
to young people, we often focus on hardware and
software. However, children today will go on to interact
WHAT IS SYSTEMS THINKING?
with, act through, and maybe even design systems in which Systems thinking looks at the overall structure of a system and the behaviour
people also play an important role. How can we prepare that emerges when its component parts interact. Instead of focusing on
young people to understand, work with, and change the lower-level details, systems thinking takes a step back and looks at an overall
systems they encounter — especially as more of these include system, often including the people who interact with it.
artificial intelligence components?
An example would be the interaction between lunch queues and timetabling in
a school. You can think of the school, with its computer systems and pupils, as
Why systems thinking?
an overall system. Systems thinking might lead us to consider offsetting the
When businesses develop solutions, they often design business
timetables of different year groups in order to reduce queuing times. However,
processes that include computer systems and the people who
we’d also need to consider the other effects that change could have — would
interact with those systems. Business processes can include tasks it affect staffing, or the timing of the end of the school day?
that are fully automated by software, tasks that are managed by
software but require human input (such as filling in a form), and
tasks that are manual but reported to a computer system (such as We need humans to decide how to use all of these different
giving medication to a patient.) methods and techniques in the best possible way. This means
More and more physical tasks are being automated, such as using systems thinking — zooming out from the detail of a system,
moving items around in an Amazon warehouse. The same is likely spotting patterns, and understanding how the parts interact.
to happen with conceptual tasks that we previously assumed This is especially necessary when considering how to use AI.
to require human intelligence. AI systems can be used to solve When we apply AI to problems, it’s important to think of the overall
certain kinds of task. impact this will have on society, jobs, and individuals. Systems
thinking is required, to consider the overall implications of a decision
or solution. We must take care not to apply AI in ways that are
locally optimal but have undesired wider implications for society.
“Unintentional destruction is [to] assume the AI wakes up
tomorrow and says oxygen is rusting my circuits … so it would …
reduce oxygen. We are collateral damage in that … they are not
really concerned, just like we are not really concerned with the
insects that we kill when we spray our fields.” — Mo Gawdat,
former chief business officer of Google X (helloworld.cc/Gawdat).

How do we teach young people to be responsible,

empowered creators?
To prepare young people for the future of work, educators can
TRACY GARDNER & MICHAEL CONTERIO provide opportunities for them to experience authentic scenarios
Dr Tracy Gardner is a computer scientist, tech industry professional, that involve impacts beyond what appears on a computer screen.
and technology educator and is currently head of technology at By making these relatable, you can help motivate young people to
Code First Girls. Michael is a former physicist and now works as an see a future for themselves in computing — as well as developing
online course production manager at the Raspberry Pi Foundation. their critical thinking skills to help them to evaluate how to use tech
and AI safely and usefully.

84 helloworld.cc
 REVIEWS BOOKS

TEACH DATA LITERACY:

A GUIDE FOR PRIMARY TEACHERS
Clear guidance backed by excellent resources to help teachers raise
the profile of an undervalued digital skill

IN FO BY Data Education in Schools | PUBLISHER n/a | PRICE free | ISBN n/a | URL helloworld.cc/TeachDataLiteracy Ben Hall

N
o, there’s not a typo in the importance of understanding data in must be given
title of this publication — it the real world. The resources are for including
should be ‘data’, not ‘digital’ literacy. based on the Problem, Plan, Data, QR code links,
The importance of digital literacy for Analysis, and Conclusion (PPDAC) making the
primary-age children has long been cycle, which statisticians R. J. resources very
recognised, and efforts to ensure MacKay and R. W. Oldford created in easy to find
our young people are equipped for 2000. The cycle breaks down the and access.
a digital world are well-established data process, from the initial This guide
and ongoing. But in a world where question to drawing conclusions, is a fantastic
so much of what we do is monitored and explains each stage in easy-to- resource for all
and analysed, do we place enough digest steps. The team also provides primary-school
importance on data literacy? practical advice and example teachers, not just those responsible
questions for teachers. for computing. It offers a path to
What is data literacy? raising the profile of data literacy and
This guide defines data literacy Comprehensive resources developing these essential life skills
as a ‘set of skills and concepts The exploration of the model is then in children. It would also serve as a
which people need to understand, complemented by a comprehensive strong foundation for professional
interpret, and make decisions set of resources drawn from a development. I can’t recommend it
based on the data they encounter variety of sources. Extra kudos highly enough.
in the world around them.’ These
skills and concepts are routinely
taught to younger children in a
variety of subjects, including maths, BEBRAS PUZZLE SOLUTION (PAGE 75)
computing, and science. The skills
associated with digital literacy, Below is a solution to the OUCC Pig Latin problem, written in Python, that scores
such as questioning, analysing, and full marks:
concluding, are cross-curricular, but
data literacy is rarely considered a
concept in its own right.
The Data Education in Schools
team from The University of
Edinburgh has been a leading
advocate for increasing the profile of
data literacy in education in Scotland.
The team has developed a
comprehensive set of curriculum
resources that emphasise the

helloworld.cc 85
 “HELLO, WORLD!” Everything you need to know about our computing and
digital making magazine for educators

Q WHAT IS HELLO WORLD? Q WHO MAKES

Hello World is a magazine and accompanying
HELLO WORLD?
A
A
podcast for computing and digital making Hello World is the official magazine of the
educators. Written by educators, for educators, the Raspberry Pi Foundation.
magazine is designed as a platform to help you find
inspiration, share experiences, and learn from each other.

Q WHY DO WE MAKE IT?

There’s growing momentum behind the idea of
A putting computing and digital making at the heart
of modern education, and we feel there’s a need to do
more to connect with and support educators, both inside
and outside the classroom.

Q WHEN IS IT AVAILABLE?
Your magazine is available three times per year.
A Check out our podcast (helloworld.cc/podcast)
and monthly newsletter (helloworld.cc/newsletter) to get
more great content between issues.

IT’S FREE!
Hello World is free now and forever as a Creative
Commons PDF download. You can download every
issue from helloworld.cc. Visit the site to see if
you’re entitled to a free print edition, too.

86 helloworld.cc
 WANT
TO GET
INVOLVED?
There are numerous ways for you to get involved with the magazine —
here are just a handful of ideas to get you started

Give us feedback Tell us your story

Help us make your magazine better — Have you had a success (or failure) you think the
your feedback is greatly appreciated. community would benefit from hearing about?

Ask us a question Write for the magazine

Do you have a question you’d Do you have an interesting article idea? Visit
like to share? We’ll feature your thoughts helloworld.cc/writeforus to submit your idea.
and ideas.

GET IN TOUCH
Want to talk? You can reach us at:
[email protected]

FIND US ONLINE NEXT

www.helloworld.cc
ISSUE OUT
JANUARY 2024
@HelloWorld_Edu THEME: COMPUTING
fb.com/HelloWorldEduMag EDUCATION AROUND
THE WORLD

helloworld.cc 87
 helloworld.cc

88 (helloworld.cc)