GLOBAL INFORMATION SOCIETY WATCH 2019

Artificial intelligence: GLOBAL INFORMATION

GLOBAL INFORMATION SOCIETY WATCH 2019

Human rights, social justice SOCIETY WATCH 2019
and development
Artificial intelligence:
Artificial intelligence (AI) is now receiving unprecedented global atten-
tion as it finds widespread practical application in multiple spheres of
Human rights, social justice and development
activity. But what are the human rights, social justice and development
implications of AI when used in areas such as health, education and
social services, or in building “smart cities”? How does algorithmic
decision making impact on marginalised people and the poor?

This edition of Global Information Society Watch (GISWatch) provides

a perspective from the global South on the application of AI to our
everyday lives. It includes 40 country reports from countries as diverse
as Benin, Argentina, India, Russia and Ukraine, as well as three regional
reports. These are framed by eight thematic reports dealing with topics
such as data governance, food sovereignty, AI in the workplace, and
so-called “killer robots”.

While pointing to the positive use of AI to enable rights in ways that

were not easily possible before, this edition of GISWatch highlights the
real threats that we need to pay attention to if we are going to build
an AI-embedded future that enables human dignity.

GLOBAL INFORMATION SOCIETY WATCH

2019 Report
www.GISWatch.org

AssociAtion for Progressive communicAtions (APc),

A rticle 19, And swedish internAtionAl develoPment cooPerAtion Agency (sidA)
Global Information Society Watch
2019
Global Information Society Watch 2019
Artificial intelligence: Human rights, social justice and development

Operational team
Valeria Betancourt (APC)
Alan Finlay (APC)
Mallory Knodel (ARTICLE 19)
Vidushi Marda (ARTICLE 19)
Maja Romano (APC)

Project coordination team

Valeria Betancourt (APC)
Cathy Chen (APC)
Flavia Fascendini (APC)
Alan Finlay (APC)
Mallory Knodel (ARTICLE 19)
Vidushi Marda (ARTICLE 19)
Leila Nachawati (APC)
Lori Nordstrom (APC)
Maja Romano (APC)

GISWatch 2019 advisory committee

Namita Aavriti (APC)
Rasha Abdul Rahim (Amnesty International)
Alex Comninos (Research ICT Africa)
Malavika Jayaram (Digital Asia Hub)
J. Carlos Lara (Derechos Digitales - América Latina)
Joy Liddicoat (Centre for Law and Emerging Technologies, University of Otago)
Andrew Lowenthal (EngageMedia)
Micaela Mantegna (Geekylegal/Machine Intelligence Lab, Center for Technology and Society, San Andres University)
Valeria Milanes (Asociación por los Derechos Civiles)

Project coordinator
Maja Romano (APC)

Editor
Alan Finlay (APC)

Assistant editor and proofreading

Lori Nordstrom (APC)

Publication production support

Cathy Chen (APC)

Graphic design
Monocromo

Cover illustration
Matías Bervejillo

We would like to extend a special note of thanks to a number of authors who have made ad honorem contributions to this edition of GISWatch.
We gratefully acknowledge the following:
Philip Dawson and Grace Abuhamad (Element AI)
Anita Gurumurthy and Nandini Chami (IT for Change)
Rasha Abdul Rahim (Amnesty International)

APC would like to thank the Swedish International Development Cooperation Agency (Sida) and ARTICLE 19
for their support for Global Information Society Watch 2019.
Published by APC
2019
Printed in USA
Creative Commons Attribution 4.0 International (CC BY 4.0)
https://creativecommons.org/licenses/by/4.0/
Some rights reserved.
Global Information Society Watch 2019 web and e-book
ISBN 978-92-95113-13-8
APC Serial: APC-201910-CIPP-R-EN-DIGITAL-302
Disclaimer: The views expressed herein do not necessarily represent those of Sida, ARTICLE 19, APC or its members.
 ETHIOPIA
THE THRIVING AI LANDSCAPE IN ETHIOPIA: ITS IMPLICATIONS
FOR HUMAN RIGHTS, SOCIAL JUSTICE AND DEVELOPMENT

Internet Society – Ethiopia Chapter (under formation) contexts.2 The AI currently in use across the world
Abebe Chekol is broadly categorised as “narrow AI”. Narrow AI
[email protected] involves single-task applications for uses such as
image recognition, language translation and au-
tonomous vehicles. In contrast, what is known as
Introduction “artificial general intelligence” refers to systems
Ethiopia, with a population of over 105 million peo- that exhibit intelligent behaviour across a range
ple (2017), is the second most populous nation in of cognitive tasks that are not anticipated to be
Africa after Nigeria, and the fastest growing econ- achieved for at least decades.3
omy in the region. It is, however, also one of the AI is creating an increasing range of new ser-
poorest countries, with a per capita income of USD vices, products and value-adds in various sectors.
783, while aiming to reach lower-middle-income AI applications can perform a wide number of in-
status by 2025. This entails significant investment telligent behaviours: optimisation (e.g. in supply
in, among others, energy, transport, quality basic chains); pattern recognition and detection (e.g. fa-
services, and accelerating agro-based industriali- cial recognition); prediction and hypothesis testing
sation by expanding the role of the private sector. (e.g. predicting disease outbreaks); natural lan-
To this effect, the role of information and communi- guage processing; and machine translation.4
cations technologies (ICTs) is important. However, For AI to take root, the digital components sup-
access to ICTs is one of the lowest, with 37.2% pen- porting the AI landscape such as the internet of things
etration of mobile-cellular subscriptions, and 13.9% (IoT), cloud computing, broadband and connectivity,
and 18.6% penetration of mobile broadband and and big data need to be developed in countries like
individuals using the internet, respectively.1 Ethiopia. While there is growing potential for large
Despite the lack of an enabling environment, data sets from commerce, social media, science and
the artificial intelligence (AI) sector has been a other sources to become available, Ethiopia, like
growing phenomenon in Ethiopia over the last five most other African countries, has the lowest aver-
years, with various reports indicating the country is age level of statistical capacity. The lack of data, or
becoming a thriving centre for AI research and de- faulty data, severely limits the efficacy of AI systems.
velopment including robotics. This is evident from In this regard, the government of Ethiopia has recog-
various initiatives happening, ranging from AI-ena- nised the importance of data and commissioned the
bled applications and services to the development drafting of the National Open Data Policy for the Gov-
of AI-powered robots. While there are promising ernment of Ethiopia5 in January 2018, which is yet to
signs of interest in AI development towards posi- be approved by the parliament.
tively contributing to socioeconomic development, Ethiopia has also recognised privacy throughout
there has also been evidence of using the capabil- its constitutional history. The most comprehensive
ities of AI adversely to violate human rights. In this privacy safeguards statement was introduced in
context, this report tries to explore the thriving AI the constitution of 19956 which protects the priva-
landscape in Ethiopia to identify its positive contri-
bution to building a better society, as well as how it
2 Smith, M., & Neupane, S. (2018). Artificial intelligence and human
adversely affects privacy, data protection and social development: Toward a research agenda. Ottawa: IDRC. https://
justice issues. idl-bnc-idrc.dspacedirect.org/handle/10625/56949
3 Access Now. (2018). Human Rights in the Age of Artificial
Intelligence. https://www.accessnow.org/cms/assets/
Setting the scene uploads/2018/11/AI-and-Human-Rights.pdf
Being an area of computer science, AI is devoted 4 Ibid.
to developing systems that can be taught or learn 5 Ministry of Communication and Information Technology (MCIT).
(2018). Consultation on the Recommendations and Working
to make decisions and predictions within specific Text of the National Open Data Policy of the Government of
Ethiopia. Addis Ababa: MCIT. www.mcit.gov.et/web/guest/-/
1 https://www.itu.int/net4/itu-d/icteye/CountryProfileReport. draft-open-data-policy-and-guideline
aspx?countryID=77 6 https://www.wipo.int/edocs/lexdocs/laws/en/et/et007en.pdf

128 / Global Information Society Watch

cy of persons, their homes and correspondence in its involvement in more than half of the software
a detailed manner. However, this has been under- programming of Sophia.11 Furthermore, one of its
mined in the past almost two decades due to the flagship projects is Solve IT12 (a pun on “solve it”)
introduction of some unfriendly laws such as the which is a nationwide competition that runs for sev-
anti-terrorism law, the freedom of mass media and en months each year. It includes teaching young
information law, and the computer crime proclama- Ethiopians about computer coding and IT hard-
tion. Since April 2018, however, with the change of ware and entrepreneurship, and challenges them
leadership in the Ethiopian People’s Revolutionary to find technology-based solutions to community
Democratic Front (the ruling party) and the new ad- problems. Organised by the US Embassy in Addis
ministration of Prime Minister Abiy Ahmed, many Ababa in partnership with iCog Labs and Humani-
positive reforms, including lifting bans on some ty+,13 Solve IT showcases the work of enthusiastic
media outlets and unblocking over 250 websites, young inventors, and the potential of using technol-
together with initiatives to review and revise the ogy to creatively solve pressing social challenges
legislation that led to these restrictions, have been faced by vulnerable groups and communities.14
seen. Other labs are blooming in the country, laying a
foundation for AI developers to develop, test and in-
Seizing the AI opportunity in Ethiopia cubate ideas for products and services that address
and its implications real community needs. For example, EthioCloud15
Ethiopia is one of the few African countries to de- allows AI developers to work in Ethiopia’s native
ploy AI solutions at scale. It started small, unlike Amharic language, creating advanced Amharic pro-
countries in Africa that attracted global technology gramming code. It runs on Microsoft’s .NET and C#
giants such as Google’s AI research lab in Accra, platforms, and converts Amharic paper documents
Ghana, IBM’s AI-oriented research labs in Kenya into editable text, and includes an Amharic text-to-
and South Africa, and Facebook’s African technolo- speech conversion system and Amharic translator.
gy hub in Lagos, Nigeria. However, young Ethiopian There are also other hubs engaged in one way or an-
AI pioneers are still making their mark in global other in AI-related activities, including iceaddis16 and
projects.7 One such initiative is the iCog Labs Soft- blueMoon,17 and other established technology-led
ware Consultancy,8 which is an Addis Ababa-based ventures such as Gebeya,18 an online marketplace for
research and development company collaborating young talent in the IT sector. Ethio Robo Robotics19
with international AI research groups and provid- is another recent AI initiative in Ethiopia. It aims to
ing services to customers around the world. The transform access to robotics training in the country
core speciality of iCog Labs includes machine learn- by focusing on children to promote the early adop-
ing-based data analytics, computational linguistics, tion of AI technologies. It works in partnership with
computer vision, mobile robots and cognitive robot- VEX Robotics,20 a US-based company whose mission
ics, and cognitive architecture, and it has a vision is to create tools that educators and mentors can use
that looks towards the widespread introduction of to shape the learners of today into the problem-solv-
artificial general intelligence. ing leaders of tomorrow.
iCog Labs was launched in 2013 with USD 50,000 All these developments in the AI sector are met
capital and four programmers, including the found-
er and chief executive officer of SinguarityNET,9 a
11 Sophia is a social humanoid robot developed by Hong Kong-
global AI marketplace, and chief scientist of Han- based company Hanson Robotics. It is powered by artificial AI
son Robotics, the Hong Kong-based engineering and capable of over 60 different facial mechanisms to create
and robotics company known for its development natural-looking expressions. It was activated and made its first
public appearance in March 2016 with the ability to display more
of the interactive humanoid robot “Sophia”.10 One than 50 facial expressions. Since then Sophia has been covered by
of the achievements of this ambitious company is media around the world and has participated in many high-profile
interviews. https://en.wikipedia.org/wiki/Sophia_(robot)
12 https://icog-labs.com/solveit
7 Gadzala, A. (2018). Coming to Life: Artificial Intelligence in Africa. 13 https://humanityplus.org
Washington: Atlantic Council. https://www.atlanticcouncil.org/images/
14 Abdu, B. (2018, 12 October). Optimism amid challenges for
publications/Coming-to-Life-Artificial-Intelligence-in-Africa.pdf
IT innovators in Ethiopia. iCog Labs. https://icog-labs.com/
8 https://icog-labs.com optimism-amid-challenges-for-it-innovators-in-ethiopia
9 SingularityNET is a global decentralised AI network that lets 15 https://www.ethiocloud.com
anyone create, share and monetise AI services at scale. https://
16 http://www.iceaddis.com
singularitynet.io
17 https://www.bluemoonethiopia.com
10 Lewton, T. (2018, 13 June). Futurists in Ethiopia are betting on
artificial intelligence to drive development. Quartz. https:// 18 https://www.gebeya.com
qz.com/africa/1301231/ethiopias-futurists-want-artificial- 19 https://ethioroborobotics.com
intelligence-to-drive-the-countrys-development 20 https://www.vexrobotics.com

ETHIOPIA / 129
with what has been described as a minimal interest tunities and enhance productivity, given the high
in investing in innovative ideas by the Ethiopian pri- cost of retooling workers for the future world of
vate sector. The young innovators complain that local work, if the outcome is not mass unemployment, it
investors would prefer to build an asset than invest in is likely to be rising inequality. Given the projected
innovation. On the positive side, the government has effect of automation on jobs26 and Ethiopia’s vision
invested 87 million euros21 in a technology park called of becoming a lower-middle-income economy by
Ethio ICT Village22 with the ambition of it becoming a 2025, there is a need to address the impact of au-
centre of excellence for scientific and technological tomation through education policy, especially now
research. Furthermore, the government has also giv- while the policy27 itself is currently under revision.
en priority attention to ICT and innovation driving its In addition to the potential risk of job losses
transformation agenda, through, for example, impos- through the adoption of AI in Ethiopia, there are
ing a quota requiring 70% of students in universities also concerns about privacy and personal data
to study in the fields of science, technology, engineer- protection. While the CEO of iCog Labs, who was
ing and mathematics (STEM). At least two universities interviewed for this report, confirmed that they had
have devoted themselves to the field of AI. Meanwhile used videos available freely online in their devel-
the Artificial Intelligence and Robotics Center of Ex- opment of the various expressions they developed
cellence,23 promoted by the ministry of science and for the Sophia robot, it does point to the need for
technology and established under the aegis of the AI practitioners to respect privacy and use data re-
Addis Ababa Science and Technology University, has sponsibly. Ethiopia faces challenges that threaten
been set up to create a close collaboration between privacy and data protection like most other African
academia and industry in the fields of AI and robotics. countries. Among others, one of these threats is the
It is well acknowledged that AI has a tremen- absence of adequate legal, regulatory and policy
dous impact on economies and businesses and has frameworks given the collection of large amounts of
the potential to revolutionise societies. However, as personal data by government entities.28
with any scientific or technological advancement, For example, the Proclamation on the Registration
there is a real risk that the use of new tools by of Vital Events and National Identity Card29 allows the
states or corporations will have a negative impact collection of personal data and the transfer of this data
on human rights.24 With the Ethiopian industrial to various institutions including intelligence authorities
parks and integrated agro-industrial parks prolifer- without the consent of data subjects. Without regula-
ating across the country, the Ethiopian government tory safeguards, the law also authorises the storage
aims to enable the manufacturing sector to contrib- of sensitive data in a central database. Likewise, the
ute to 20% of Ethiopia’s GDP and 50% of the export incumbent telecom operator, Ethio Telecom, collects a
volume by 2025. The question is, with the industrial lot of personal information to register SIM cards. A cus-
application of AI gaining momentum, what will be tomer needs to provide detailed information including
the scale of industrial job losses due to automation? name and address, a photo ID, a photograph, and a sig-
According to World Bank Development Report nature before one can purchase a SIM card.30 Another
2016 estimates, two-thirds of all jobs are suscepti- concern is the use of surveillance technologies by gov-
ble to automation in the developing world, and the ernment law enforcement agencies to gather personal
share of jobs at risk of being lost to automation and data without putting in place regulatory mechanisms
advanced technologies is about 85% for Ethiopia.25 to protect personal data.
The implication is that in the absence of adequate
policies, many workers are likely to be pushed into 26 Ibid.
lower-wage jobs or become unemployed, even if 27 Teferra, T., et al. (2018). Ethiopian Education Development
temporarily. While technologies create new oppor- Roadmap (2018-30): An Integrated Executive Summary – Draft
for Discussion. Addis Ababa: Ministry of Education, Education
Strategy Center. https://planipolis.iiep.unesco.org/sites/
planipolis/files/ressources/ethiopia_education_development_
21 Karas-Delcourt, M. (2016, 28 January). The Ethiopian AI Geeks roadmap_2018-2030.pdf
Building Cutting-Edge Robots. iCog Labs. https://icog-labs.com/ 28 Yilma, K. M. (2015). Data privacy law and practice in Ethiopia.
the-ethiopian-ai-geeks-building-cutting-edge-robots/#more-1295 International Data Privacy Law, 5(3), 177-189; see also Enyew, A. B.
22 ethioictvillage.gov.et/index.php/eng (2016). Towards Data Protection Law in Ethiopia, in A. B. Makulilo
23 www.aastu.edu.et/research-and-technology-transfer-vpresident/ (Ed.), African Data Privacy Laws. Springer International Publishing.
the-artificial-intelligence-robotics-center-of-excellence 29 Federal Democratic Republic of Ethiopia. (2012). A Proclamation
24 Privacy International & ARTICLE 19. (2018). Privacy on the Registration of Vital Events and National Identity Card,
and Freedom of Expression in the Age of Artificial Proclamation No. 760/2012. https://chilot.files.wordpress.
Intelligence. https://privacyinternational.org/report/1752/ com/2013/04/proclamation-no-760-2012-registration-of-vital-
privacy-and-freedom-expression-age-artificial-intelligence events-and-national-identity-card-proclamation.pdf
25 World Bank Group. (2016). World Development Report 2016: Digital 30 Taye, B., & Teshome, R. (2018). Privacy and Personal Data
Dividends. https://www.worldbank.org/en/publication/wdr2016 Protection in Ethiopia. CIPESA. https://cipesa.org/?wpfb_dl=301

130 / Global Information Society Watch

 Furthermore, AI also has the potential to impact a positive and negative role to protect and to refrain
negatively on freedom of expression. Using irre- from interfering in the citizens’ exercise of their rights
sponsible social media activism and fake news that and freedoms. In many ways, the internet – and par-
has recently catalysed ethnic tension and violence ticularly emerging technologies – have opened ways
in the country as a pretext, the government has pro- for the exercise of many rights and freedoms while
posed to pass a new law on hate speech.31 Although at the same time challenging them. Examples in-
hate speech is a growing concern, it also has to be clude online hate speech, fake news, surveillance,
handled with care given the potential of new tech- or privacy issues. In this regard, the main role of the
nologies such as AI to manipulate video, audio and government is to ensure the balance between free-
images. For example, “deepfake” technology uses dom and protection, rights and responsibilities. This
machine learning to help users edit videos and add, maintenance of the right balance requires the gov-
delete or change the words coming right out of ernment to engage the private sector, civil society
somebody’s mouth.32 Such emerging technologies and other stakeholders in respecting human rights,
can exacerbate the potential risk of AI and its im- including in the design, development and delivery of
plications for human rights and social justice if not its own digital services.33
responsibly used by the public, corporations, the
state and other stakeholders. Action steps
Ethiopia can reap the benefits of AI if the Ethiopian gov-
Conclusion ernment, investors and other stakeholders can equip
Ethiopia’s AI landscape is surrounded by both opti- workers with 21st century skills, and reform laws and
mism and fear; optimism as to the potential that AI education to meet the demands of the digital economy.
has for economic and social development, and fear To this end, the following actions are proposed:
of its human rights implications. While increasing • Ensure that the education development road-
youth enrolment in STEM fields and supporting AI map currently under discussion is forward-look-
innovation brings about economic benefits through ing so that it integrates AI studies in the
the creation of new job streams, AI also has nega- educational system and meets the demands of
tive consequences with regard to the susceptibility tomorrow’s economy.
of the country to job losses due to automation. Giv-
• Review the national ICT policy so that it embrac-
en the impact on future jobs, there will be demand
es new emerging technologies including AI, big
for retooling, and its associated cost will contribute
data, IoT and cloud computing.
to rising inequality. While the government is accel-
erating agriculture-led industrialisation through • Put in place legal frameworks with respect to
establishing industrial parks across the country, the privacy and data protection, taking into account
anticipated impact on unemployment is likely to be the African Union Convention on Cyber Security
reduced given the take-up of automation in the man- and Personal Data Protection.34
ufacturing industries. With unemployment ranging • Build a sound statistical system that adapts to
at different times between 16% and 26%, the neg- the emerging data revolution.
ative impact of AI on jobs would be significant. • Promote the thriving innovation labs so that
Furthermore, the increasing rate of unemployment they can leverage innovations across industrial
in Ethiopia, particularly among young, college-ed- parks in the country.
ucated people, presents a challenge to economic
• Manage the industrialisation process in order to
development, placing the onus on the government
ensure gender-responsive outcomes that ben-
and universities to develop future-ready skills for
efit women and girls,35 including society as a
the technology sector.
whole, through the removal of barriers to equity.
Governments have the main responsibility to pro-
tect human rights. They act as the primary guarantors
of these rights, and should be held accountable when 33 DiploFoundation. (2018). Mapping the challenges and
rights are not realised. In this regard, they have both opportunities of artificial intelligence for the conduct of diplomacy.
Diplo AI Lab and Ministry of Foreign Affairs of Finland. https://
www.diplomacy.edu/AI-diplo-report
31 Tsegaye, Y. (2018, 23 November). Ethiopia Preparing New Bill to 34 https://au.int/sites/default/files/treaties/29560-treaty-0048_-_
Curb Hate Speech. Addis Standard. https://addisstandard.com/ african_union_convention_on_cyber_security_and_personal_
news-ethiopia-preparing-new-bill-to-curb-hate-speech data_protection_e.pdf
32 Vincent, J. (2019, 10 June) AI deepfakes are now as simple 35 United Nations Development Programme. (2018). Ethiopia National
as typing whatever you want your subject to say. The Human Development Report 2018: Industrialization with a Human
Verge. https://www.theverge.com/2019/6/10/18659432/ Face. Addis Ababa: UNDP. hdr.undp.org/sites/default/files/
deepfake-ai-fakes-tech-edit-video-by-typing-new-words ethiopia_national_human_development_report_2018.pdf

ETHIOPIA / 131
GLOBAL INFORMATION SOCIETY WATCH 2019
Artificial intelligence: GLOBAL INFORMATION

GLOBAL INFORMATION SOCIETY WATCH 2019

Human rights, social justice SOCIETY WATCH 2019
and development
Artificial intelligence:
Artificial intelligence (AI) is now receiving unprecedented global atten-
tion as it finds widespread practical application in multiple spheres of
Human rights, social justice and development
activity. But what are the human rights, social justice and development
implications of AI when used in areas such as health, education and
social services, or in building “smart cities”? How does algorithmic
decision making impact on marginalised people and the poor?

This edition of Global Information Society Watch (GISWatch) provides

a perspective from the global South on the application of AI to our
everyday lives. It includes 40 country reports from countries as diverse
as Benin, Argentina, India, Russia and Ukraine, as well as three regional
reports. These are framed by eight thematic reports dealing with topics
such as data governance, food sovereignty, AI in the workplace, and
so-called “killer robots”.

While pointing to the positive use of AI to enable rights in ways that

were not easily possible before, this edition of GISWatch highlights the
real threats that we need to pay attention to if we are going to build
an AI-embedded future that enables human dignity.

GLOBAL INFORMATION SOCIETY WATCH

2019 Report
www.GISWatch.org

AssociAtion for Progressive communicAtions (APc),

A rticle 19, And swedish internAtionAl develoPment cooPerAtion Agency (sidA)