Brief Guide

Learning and Teaching

with AI tools
This guide is intended to introduce Artificial Intelligence (AI),
specifically Generative Artificial Intelligence (GenAI) and its
implications for teaching and learning. AI technologies are
developing rapidly, and this guide will be updated
occasionally to incorporate these developments.
ChatGPT dominated headlines with the release of ChatGPT
3.5 in November 2022 and has resulted in an explosion of
GenAI services that create text, images, code, video and
even voices during 2023. These will likely grow and
continue to develop new capabilities during 2024.

This guide aims to answer the following:

● What is GenAI and how does it work?
● What are the current capacities and limitations?
● What are the current applications in teaching and
learning?
What is AI?
Artificial Intelligence (AI) is a technology that has been around since the 1950s,
meaning that it is not as new as we think. It's aim is developing intelligent machines
that mimic human behaviour and which can learn from its interaction with the world.
This includes various technologies such as machine learning, natural language
processing, data mining, neural networks and algorithms. It is used in many software
solutions we use in devices, such as predictive text, location-based services, etc. This
article gives a good introduction to what AI is and this article gives some explanation
of the key technologies.

What is Generative AI, and what is the fuss?

Generative Artificial Intelligence (GenAI) is a specific form of AI that rapidly generates
text, images, audio or video. ChatGPT is one kind of GenAI, but there are a variety of
other AI services. ChatGPT, in its current form, was released in November 2022 and
disrupted the notion that writing was exclusively a human activity. The disruptive effect
soon became a central concern for many in society, especially the education system,
because GenAI can create texts and essays that can lead to academic dishonesty.
Apart from ChatGPT there are a variety of other AI services like Google Bard with its
newly launched Google Gemini and Claude. The rise of these GenAIs led to the
development of other services like Elicit, Litmaps and SciSpace, which can find, read
and summarise literature instantly. Most of these are free to use with advanced
functionalities and models requiring a paid-for account. These GenAI models
continuously evolve and develop, and the latest version of ChatGPT Plus (the paid
version) can now “see, hear and talk” via the web service or the App. It can interpret
graphs uploaded and understand physics in an image. ChatGPT’s abilities are further
enhanced by plugins that, for example, connect to academic databases and aid in
literature searches by searching for literature. Some plugins also offer ways to assist
reading by explaining text or providing a way to use data analytics using natural
language prompts. For example, Google Bard can access YouTube videos and it can
answer questions about the videos. The boundaries of what is possible will
continuously shift as technologies evolve and develop new capabilities like ‘seeing,
hearing and listening’ as well as completing more tasks. Many of the affordances and
challenges these bring are only evident when enough time has passed or if we use
them actively.
What is ChatGPT?
• ChatGPT is a GenAI
• GPT stands for ‘Generative Pre-trained Transformer’, meaning it is pre-trained to
answer questions in a particular way based on a database of information (not
always to the Internet) and not to answer harmful questions - something it does
not always do well.
• OpenAI released ChatGPT 3.5 in November 2022 as a free-to-use research
version but has an improved ChatGPT Plus, version 4, as a paid service.
• ChatGPT works like a chatbot powered by a Large Language Model (LLM) that
performs natural language processing tasks, like answering questions, creating
text or classifying text, and translating text in a conversational style. This article
explains this well.

How does ChatGPT and other GenAIs work?

Most GenAI tools work in a similar way to ChatGPT. ChatGPT responds to prompts
(questions or statements) using a natural language processing algorithm and the large
language model database it is trained on. When ChatGPT is asked a question, the
following happens:
1. The algorithm analyses the prompt (question) using the LLM database ChatGPT
is trained on.
2. The algorithm then generates an output (response) using the database to
construct a response.
3. The response is generated using probability logic. This means it creates a
response based on the probability of the next word in a string of words to create
a sentence that makes sense for a human reader. For example – if we ask
ChatGPT to ‘tell me a story about a cat’.

● ChatGPT notes the word 'story' and will draw on its massive data set to
generate a text that follows the most common structure used in texts
identifiable as 'stories'.
● ChatGPT notes the word 'cat' and will draw on its massive data set to generate
a text that relates to the most common texts on 'cats'.
● ChatGPT uses these almost instant searches to present a text that is a 'story'
about a 'cat'.
● This story can be refined with prompts like “Retell this same story but make it
suitable for children under 5 years of age” and the output would be adapted.

4. If ChatGPT does not have any information about a question it is asked:

● It will create a response to complete the sentence, which is known as a
‘hallucination’.
 ● ChatGPT is trained to use its large data set to create texts that are meaningful,
sound and convincing to humans rather than create texts that are true.
● The database on which ChatGPT is trained may contain prejudices and
inaccurate data used to create its responses.
● For example, if you ask ChatGPT for literature, it may, in some instances,
create fake references which do not exist.

GenAI and education

The disruptive effects of GenAIs is a concern for many in the education system as it
allows students to use AI-generated text in essays or examinations. The ease of
access to these powerful tools which can write human-like essays has led to a focus
on academic dishonesty. For others, GenAI has the potential to change teaching and
learning fundamentally in both positive and negative ways.
What is certain is that these new technologies are ever-evolving and will become
ubiquitous. As academics, we need to experiment with these technologies in different
disciplines to understand the implications. Incorporating AI technology into teaching
and learning needs to be undertaken to:
1. Expose our students to emerging technologies in a controlled environment to
build their critical understanding of the technologies in various disciplines.
2. Develop an understanding of the influence of AI technologies on knowledge-
building in our disciplines.
3. Gain a clear understanding of the AI technologies available and the affordances
and challenges these technologies bring.

Risks and ethics of using GenAI

ChatGPT is a useful tool that can be used for learning. But you have to know and be
aware of the following:
● Database inaccuracies - ChatGPT creates convincing outputs/responses that
sound reliable, convincing, and factually accurate. However, the software does
not understand the content and cannot distinguish between correct or incorrect
information. ChatGPT can ‘hallucinate’ and sound authoritative on many topics.
Therefore, you need to evaluate and verify all responses from ChatGPT and
other AIs to verify the information generated.
● Bias - AIs are trained from internet data. The information on the internet, at its
core, is biased in explicit and implicit ways. The data favours Western knowledge
and ignores localised knowledge with different forms of sexism, racism, ableism
and other negative ideologies. This means ChatGPT has the potential to replicate
the biases from the database it is trained on. This means we need to be aware
of the potential for bias in the outputs/responses and evaluate all the information
produced critically. This article explains bias in AI well.
● Reliance on, and the use of GenAI - In the future, GenAI, like ChatGPT, will
likely play a significant role in how we learn and practice new skills and assist
 with some routine tasks. However, it is still important we develop the critical
thinking capabilities of our students and develop their cognitive capabilities and
introduce them to the knowledge in our disciplines.
● Privacy and ethical use – We are bombarded with terms and conditions for
every device, app and service. Most of us simply click accept and move on to
use the service. ChatGPT and other GenAIs are no different. We need to know
what information is used and shared and who owns the prompts and output from
the different GenAIs. The content we contribute to these services holds
significance, and we need to know how the information we upload on these
platforms is handled. For example, using GenAI tools to transcribe audio files
may result in inadvertently releasing confidential research materials to the public
domain.

How could we approach the use of GenAI technologies in

teaching and learning?
The use of GenAI has been debated frequently in academic literature, and research
on how to integrate AI into teaching and learning is emerging. One approach is the
RITE approach, which emphasises responsible, informed, transparent and ethical use
of AI and provides a useful framework on how to incorporate AI in teaching and
learning to foster positive use of AIs.
Be responsible: focus any teaching and learning activity on what responsible use of
GenAIs should be. For example, to brainstorm ideas, overcome writer's block or make
sense of content but importantly to critically evaluate all outputs. Responsible use is
centred around how the tool can enhance understanding and learning and assist in
developing a student’s writing style, improve their thinking and knowledge.
Be informed: design teaching and learning activities to foster a deep understanding
of the limitations and risks of using GenAIs. This includes issues like bias,
hallucinations and other practical concerns around the quality of the outputs to foster
AI literacies. Being informed means understanding who makes these models and for
what purpose. It is being able to answer questions about the terms of use, including
data and privacy concerns. Who owns the outputs of the AI, and who owns what is put
in? What happens if any of the data is leaked? Being informed is also about knowing
the ethical debates around how the models are built and the intellectual property
infringements which may occur. For example, knowing that the Kenyan labour was
exploited to build and train these models and the data. Lastly, these tools have an
environmental impact that needs to be considered.
Be transparent: design teaching and learning activities that foster student abilities to
acknowledge the use of AI by setting out the conditions and expectations of use clearly
in course materials.
Be ethical: design teaching and learning activities that create a suitable environment
for ethical use. If the use of AI is allowed, provide students with the tools and a
framework to acknowledge use and ways to provide evidence of the use of AI. For
example, provide a means of citation, acknowledgment, or space in the submission of
the assignment for evidence of how AI has been used to produce the assignment.
Fostering critical AI literacies
GenAI will transform the world of work and education in unknown ways and will require
different skills from academics and students for a world augmented by AI. This places
a responsibility on higher education to respond to this shifting technological setting
and the changing ways of engaging with knowledge. It will be imperative to incorporate
AI literacies in our teaching and students’ learning in the same way academic literacies
are incorporated into teaching and learning.
AI literacies include:
● Knowledge about the functions of AI
● Ethical use of AI
● Ability to critically evaluate AI and its output
● Understanding what AI can and cannot be used for
● Understanding the limitations of AI
● Understanding prompt engineering - OpenAI has released a guide

AI literacies do not exist in a vacuum and require an integrated set of literacies

including academic literacies, disciplinary literacies, information literacies and critical
literacies. The ability to evaluate information and sources in a critical way using a
disciplinary lens is a key outcome of the acquisition of AI literacies.

What can we use GenAIs, like ChatGPT, for in teaching and

learning?
The table on the next page provides some examples:
Role Description Example of implementation
Possibility AI generates Students write queries in ChatGPT and use the
engine alternative ways of regenerate response function to examine
expressing an idea alternative responses.
Socratic AI acts as an Students enter prompts into ChatGPT following
opponent opponent to develop the structure of a conversation or debate.
an argument Teachers can ask students to use ChatGPT to
prepare for discussions.
Collaboration AI helps groups Working in groups, students use ChatGPT to
coach research and solve find information to complete tasks and
problems together assignments.
Guide on the AI acts as a guide to Lecturers use ChatGPT to generate content for
side navigate physical and classes/courses (e.g., discussion questions)
conceptual spaces and advice on how to support students in
learning specific concepts.
Personal tutor AI acts as a tutor for ChatGPT provides personalized feedback to
students by providing students based on information provided by
immediate feedback students or lecturers (e.g., test scores).
on progress
Co-designer AI assists throughout Lecturers ask ChatGPT for ideas about
the design process designing or updating a curriculum (e.g., rubrics
for assessment) and/or to focus on specific
goals (e.g., how to make the curriculum more
accessible).
Exploratorium AI provides tools to Lecturers provide basic information to students
play with, explore and who write different queries in ChatGPT to find
interpret data out more. ChatGPT can be used to support
language learning.
Study buddy AI helps the student Students explain their current level of
reflect on learning understanding to ChatGPT and ask for ways to
material help them study the material. ChatGPT could
also help students prepare for other tasks (e.g.,
job interviews).
Motivator AI offers games and Lecturers or students ask ChatGPT for ideas
challenges to extend about extending students’ learning after
learning providing a summary of the current level of
knowledge (e.g., quizzes, exercises).
Dynamic AI provides educators Students interact with ChatGPT in a tutorial-
assessor with a profile of each type dialogue and then ask ChatGPT to
student’s current produce a summary of their current state of
knowledge knowledge to share with their teacher/for
assessment.
Adapted from the UNESCO guides on Artificial Intelligence
Provide your students with the following guidelines for
using AI:
1. Understand the basics: Familiarize yourself with the fundamental concepts of
AI to make informed decisions about its usage.
2. Academic integrity: Use AI tools ethically and responsibly. Don't plagiarize or
use AI to generate content without proper attribution.
3. Supplement learning: Utilize AI for research, data analysis, and complex
problem-solving, but remember that it's a tool to enhance your learning, not
replace it.
4. Critical thinking: Always critically evaluate AI-generated results. Cross-
reference information from reliable sources before drawing conclusions.
5. Privacy awareness: Be cautious about sharing personal information with AI
platforms and review their privacy policies.
6. Bias awareness: Recognize that AI systems can carry biases from their
training data. Be mindful of potential bias in AI-generated content.
7. Credible sources: Rely on reputable AI tools and resources. Check reviews
and recommendations before using new platforms.
8. Learn and experiment: Take time to understand how AI tools work.
Experimenting and learning about their capabilities can enhance your skills.
9. Stay updated: AI technology evolves rapidly. Stay up to date with the latest
advancements and trends in the field.
10. Ethical AI use: Consider the broader societal implications of AI. Think about
how AI can impact various aspects of society and strive to use it responsibly.

Recommendations/Examples
Lecturers can sign up for a free version of ChatGPT (or GPT-4) on Open AI’s platform.
A first step in exploring the tool may be to enter some of your assignment prompts and
assess the accuracy of the output. Then reflect on how you might embrace the tool or
implement assessment strategies that make use of the tool unnecessary or
undesirable.
TIP: Including your students in the reflective process is also a learning opportunity to
help them understand the benefits and limitations of the tool.
A few things to try:
● Ask ChatGPT a question – it could be a homework assignment or in-class
question.
● How would you evaluate the response provided by ChatGPT?
● Try modifying the prompt and see how that changes the response.

● Ask ChatGPT to synthesize text from large documents. For example, enter
a 3500-word paper as a prompt and ask ChatGPT to create an 18-slide
PowerPoint presentation with headings and bullet points, making a persuasive
case for action.
● Prompt for writing samples specific to your area of expertise. For example,
ask ChatGPT to generate advice a pharmacist might give to a 53-year-old male
who is seeking pain medication for chest pain. Another example is to ask for an
email introducing your upcoming course to enrolled students.
● Ask ChatGPT to translate a piece of text into another language. Together
with your students examine the translation to see how well it did. Ask for
improvements, or consider when such translation capabilities might be limited.
● If you teach students how to code, ask ChatGPT to correct incorrect code
(debug code). Consider ways this might help students who are learning to code
in your course.
● If you teach a writing intensive course, try asking ChatGPT to respond to a
writing prompt in a specific style (like a popular author) or create a poem on a
specific topic.
● Reflect on the potential of ChatGPT to support student writing. Which
writing practices do you feel are fundamental for students’ appropriate
use of artificial intelligence? How might ChatGPT facilitate the
development of writing or problem-solving skills?
● Have students use ChatGPT to write a draft of an assignment and then
have them edit what it produces or check for errors