By Javier Surasky-
Artificial Intelligence
is changing our world, and debates on how to deal with this technology are happening
worldwide. The G20 members are defining the core elements of a potential shared
position: principles, approaches, and goals for a necessary but controversial AI global
governance building.
As part of the
way to the G20 Summit in Rio de Janeiro in November 2024, the Ministers responsible
for the Digital Economy of the Group’s countries met in Maceio, Brazil, in September
2024, with an Agenda focused on digital inclusion, connectivity, digital government,
digital public infrastructure, online information integrity, trust in the digital
economy, and artificial intelligence for sustainable development and inequalities
reduction.
A “G20 Maceio Ministerial
Declaration on Digital Inclusion for All” was adopted, including four annexes: the Guidelines
on Indicators and Metrics for Universal and Meaningful Connectivity, the General
Principles on the Governance of Digital Identity, and two declarations on Promoting
Information Integrity and Trust in the Digital Environment, and on Enabling resources
for the development, deployment, and use of AI for good and for all.
In this post,
I will focus on the chapter on AI for inclusive, sustainable development and inequality
reduction and the annex on the development, deployment, and use of AI for good.
I aim to explore the G20 approach to this issue, particularly when it comes to AI
international governance.
The G20 Ministers
start by recognizing that safe, secure, and trustworthy AI can be a catalyst for
achieving economic growth and inclusive and sustainable development if it is used
in a transparent, ethical, responsible, and reliable form, reaffirming the G20 AI
principles, the UNESCO Recommendation on the Ethics of AI, the G20 New Delhi Leaders’
Declaration, and its commitment to leveraging AI for good, sharing its benefits
for all and mitigating its risks.
The Ministers
indirectly support the role of the UN General Assembly as a critical scenario to
discuss AI-related issues by acknowledging the adoption of its two resolutions on
AI, namely “Seizing the opportunities
of safe, secure and trustworthy artificial intelligence systems for sustainable
development” and “Enhancing International
Cooperation on Capacity Building of Artificial Intelligence.” Moreover, they expressed their interest
in the report of the Secretary-General’s
High-Level Advisory Body on Artificial Intelligence, which was still not published at the time of the
meeting.
In that line,
the ministers committed to following a risk-based, human-centric, development-oriented,
innovation-friendly AI governance approach consistent with existing legal frameworks
on personal data security, privacy and protection, and human and intellectual property
rights, introducing a legal basis into their approach to AI governance.
A mention of
global challenges arising from the AI divide and multiple disparities within and
between developed and developing countries and a call for the promotion of international
cooperation, capacity building, joint research, and voluntary technology transfer
and knowledge sharing is made, while open-source technologies and open data are
highlighted as crucial to delivering AI benefits at scale. Although balanced by
recognizing the need to ensure AI systems respect the diverse linguistic, sociocultural,
racial, and geographical contexts to avoid reinforcing or perpetuating real-world
discrimination and bias.
Brazil's Presidency
efforts on AI are showcased by mentioning its Toolkit for Artificial
Intelligence Readiness and Capacity Assessment, produced with the support of UNESCO, and the report
“Mapping of AI Adoption
for Enhanced Public Services in the G20.” In the same way, the G20 event “Harnessing Artificial Intelligence for
Social Equity and Sustainable Development” is underlined as a chance for G20 members and
invited countries to equip governments and “relevant stakeholders,” including the
private sector, academia, and civil society, with the capabilities and tools necessary
to respond to conditions, possibilities, and capacities disparities-related challenges.
The continuous use of the word “relevant” is dangerous: even if it is broadly accepted,
it creates doubts about who is “relevant” and who will decide who is relevant in
a debate that affects everyone’s future.
Previewing the
future G20 Presidency efforts, the Ministers welcomed South Africa’s plans to further
work on AI, the deployment of digital public infrastructure, and digital innovation
ecosystems to support MSMEs and advance universal and meaningful connectivity, renewing
the ToR for the G2o Digital Economy Working Group.
Annex 4 on “Enabling
resources for the development, deployment, and use of AI for good and for all” is
introduced as part of the Minister's commitment to further discussions on AI for
sustainable development and fighting inequality.
Consequently,
the annex highlights that “if suitable guardrails are not in place, there is a risk
that AI systems could reproduce and exacerbate societal biases, economic inequalities,
and digital divides, among others.” This lack of precautions happens in the framework
of huge differences in enjoying AI opportunities among countries. The Ministers
recognized that the G20 plays a pivotal role in leading collaborative efforts to
harness the potential of AI for inclusive and sustainable development, which requires
an enabling environment, including physical and digital infrastructure, human capital,
data availability, research capabilities, and digital talent, accompanied by suitable
governance, institutional settings, and a dynamic business sector, which explains
why the Brazil Presidency has prioritized fostering partnerships and cooperation
among countries and with relevant stakeholders, among other key issues.
A short list
of pretty-obvious findings is included in the annex:
- Most G20 members and invited countries
are engaged in shaping their institutional and governance AI frameworks and
have ongoing initiatives to enhance digital skills and incorporate AI into
secondary and higher education.
- Relevant AI for Sustainable Development
deployment gaps emerge between and within countries. Innovation capacity
for AI also differs across countries.
- Only a few G20 countries have strategies
to measure total AI R&D spending. Nevertheless, investments in AI research
and innovation are taking place, and safe, secure, and trustworthy development,
deployment, and use of AI in a transparent, ethical, responsible, and reliable
manner are increasingly cornerstones of national AI strategies.
- There is still a need to implement
policies to address the gender gap and the environmental agenda relations with
AI further.
Based on those
“findings,” the Brazilian Presidency outlined three axes and recommendations to
support the AI ecosystem development.
- Promote international collaboration and multistakeholder partnership by enhancing exchanges on policies to foster interoperable frameworks to promote the safe, secure, and trustworthy development and deployment of AI systems in an ethical and responsible way while promoting economic growth, technological cooperation, and sustainable development, safeguarding human rights, fundamental freedoms, inclusiveness, equity, and accessibility.
- Enhance AI enabling environments and capability building by strengthening AI enabling environments and enhancing AI capabilities, investing in technology and technology-related infrastructure, education, and skill-building initiatives, with particular attention to low- and middle-income countries and communities, improving countries' capacities to determine their digital future.
- Develop governance
frameworks and policies by developing and implementing innovation-friendly and forward-looking governance
frameworks and policies that promote digital and AI empowerment.
The three recommendations
are explicitly linked to harnessing G20 and invited countries' capabilities “to
cultivate an AI ecosystem that supports an inclusive, sustainable, equitable growth
and distribution of benefits.”
Conclusions:
Despite the
lack of precision, the Declaration of Maceio of the G20 Ministers in charge of digital
technologies shows some group priorities.
- Besides the well-established mention of new opportunities for development arising from AI technologies and the consequent risks linked to digital technology development, the G20 repeatedly refers to “AI for inclusive sustainable development” as an explanation of what “AI for good” means in operative terms.
- Accordingly, the G20 views are closely linked to reducing digital inequalities and bridging the AI divide within and between countries. International cooperation on infrastructure, technical, or human capabilities development is identified as a major tool for reaching group goals. Sadly, this push for inclusion is not paired with a consideration of the role of non-state actors in advancing those goals.
- The AI governance approach arising from the document is risk-based, human-centric, development-oriented, innovation-friendly, and participatory. Human and property rights inform the G20 basis for AI governance.
- Interestingly, G20 Ministers mention environment protection and gender-related issues as insufficiently addressed by AI public policies so far and recognize the defense of local specificities as a way to face discrimination and bias. Still, no mention of youth or future generations is made.
- A message on the UN system as a central part of developing a shared AI governance is easy to identify, although it is not expressly stated. UNESCO is seen as a critical counterpart of the G20 efforts, getting more attention than other UN system parts with key responsibilities in the area, such as ITU.
Even when the
discussions are open and will continue beyond the 2024 G20 Summit, with South Africa
assuming the lead, an initial G20 basic consensus on how AI global governance should
look is emerging. It will be wise to pay attention to it as much as we look at the
United States, China, and the European Union's positions on AI governance building.
