Task Force Title: TF-7: Towards Reformed Multilateralism: Transforming Global Institutions and Frameworks
1. The Challenge
The disruptive impact of artificial intelligence (AI) cannot only be viewed from a techno-deterministic lens. As much as technology influences society, society also impacts technological innovations via governance, principles, technical standards, diffusion, adaptation, and integration. On the one hand, robust national systems of innovation (NSI) require effective domestic governance of science technology and innovation (STI) and coordination between the triple helix actors. On the other hand, our multidimensional transnational interdependencies highlight that international cooperation for responsible AI governance is necessary if the majority of the Global South are to catch up and compete with technologically advanced high-income countries in the Global North. This will ultimately enable the Global South to navigate the multidimensional challenges and existential risks that AI potentially presents to our interdependent global system.
The following challenges underline the need for reformed multilateralism for inclusive responsible AI global governance:
The rise of neo-techno-nationalists and global regulatory uncertainty
For decades, harmonised global technical standards, regulations, policies, and norms have enabled access to global (digital) public goods, technology, and knowledge transfers, and the deployment of cutting-edge lifesaving research. They have also facilitated the unprecedented speed of development in frontier technologies that we now associate with data-driven digital transformation. Paradoxically, these hyper-globalisation-induced processes have also negatively impacted developing countries because many STI policy instruments that have been created to improve research and development (R&D) and broader innovation ecosystems in the Global North often fail to capture the techno-socio-political complexities of innovation ecosystems in the Global South., Geopolitics and global market-driven capitalist agendas have also played a role in influencing institutions that shape global governance, often to suit techno-nationalist agendas. The competition however, for data control and tech supremacy has transitioned beyond techno-nationalism and precipitated to a neo-techno-nationalist race, spearheaded by a few geopolitical powerhouses that influence the development of the data-driven digital economy.
The AI Big Three—China, the European Union (EU) and the US—arguably shape the new global order of governance, deployment, and development of AI and the broader data-driven digital economy in support of their interests. This has led to a fragmented international regulatory regime, which lacks clear harmonised guidelines, values, and technical standards that support the developmental needs of countries in the Global South. It is evident that the AI Big Three aim not only to control and own global critical infrastructure and software, and hardware value chains that are prerequisites for national AI deployments. Their objective is also the diffusion of ideological values and technical standards to control, reshape institutions, and frame global governance of AI developments and deployments beyond their jurisdictions. At an industry level, the AI Big Three are headquarters of the top 200 most influential digital technology companies worldwide, and they shape current industry-led global AI governance with ethical AI frameworks.
Exclusionary global AI governance mechanisms and uneven power dynamics
Various global multistakeholder initiatives have raised awareness about AI governance challenges with diverse stakeholders to address the many aspects of AI governance, including tackling issues related to human rights, data governance, and innovation (see Table 1). However, the lack of formal decision-making structures, binding mechanisms, and enforceable regulations undermines the effectiveness and impact of these existing initiatives, leading to a fragmented and inconsistent landscape of global AI governance.
Table 1: The current actors and institutions in global AI governance
|Institution||Description||Role in AI Governance||Advantages||Limitations||Specialised Institution|
|United Nations||Intergovernmental organization||Facilitating international discussions on AI policies and standards||Global reach and legitimacy||Lack of binding enforcement power||
ITU (International Telecommunication Union)
Roundtable 3C on AI under the United Nations Secretary-General’s (UNSG) Digital Roadmap
|OECD (Organisation for Economic Co-operation and Development)||International economic organisation||Developing principles and guidelines for trustworthy AI||Strong research and policy expertise||Limited membership, primarily focused on developed countries||OECD AI Policy Observatory|
|G20||Forum for international economic cooperation||Addressing global challenges and opportunities of AI||Involvement of major economies||Limited ability to enforce policies across member states||N/A|
|World Economic Forum||International organisation for public-private cooperation||Advancing responsible AI development and deployment||Multi-stakeholder engagement and partnerships||Participation limited to member organisations||Global AI Council, Center for the Fourth Industrial Revolution (C4IR)|
|AI Global Governance Commission||Independent expert commission||Researching and providing recommendations on AI governance||Neutral and independent expertise||Limited direct policymaking authority||N/A|
|Partnership on AI (PAI)||Multi-stakeholder initiative||Promoting ethical AI principles and best practices||Collaboration among diverse stakeholders||Voluntary participation, non-binding commitments||N/A|
|Global Partnership for AI||International collaboration platform||Developing AI policies and fostering responsible AI innovation||Focus on AI ethics, inclusion, and human rights||Relatively new initiative, impact remains to be seen||ISO/IEC JTC 1/SC 42 (Artificial Intelligence)|
|Multinational Tech Companies||Google, Microsoft, IBM, etc.||Developing and implementing AI technologies||Cutting-edge research and development capabilities||Potential for concentration of power and influence||
IEEE Standards Association, ISO/IEC JTC 1/SC 42
Information Technology Industry Council (ITI)
|IEEE (Institute of Electrical and Electronics Engineers)||Professional association||Setting technical standards for AI||Broad technical expertise and industry participation||Primarily focuses on technical aspects of AI governance||IEEE Standards Association|
Source: Various sources
There is an insufficient emphasis on addressing representation imbalances and power asymmetries, resulting in the limited influence of the key stakeholders, including the academia, civil society, private sector, and public authorities of the Global South in shaping global AI governance. There is also a lack of consideration for the fact that Western knowledge, values, and ideas, which function well in one environment might not function as effectively when adopted elsewhere or be maintained after conditions change. This is particularly so given the exponential breakthroughs that are commonplace in general-use AI developments. 
Furthermore, many Global South actors have also argued that current transnational AI governance frameworks are exclusionary. Their expertise in interpreting AI risks has been overlooked, and existing initiatives do not adequately consider the perspectives of Global South experts. Without active measures to facilitate meaningful participation in the multidimensional global AI governance discourse, countries in the Global South will likely find it challenging to limit the harm caused by AI-based disruptions.
Lack of effective transversal governance solutions for overlapping challenges
In the rapidly evolving landscape of technological advancements, addressing interdisciplinary challenges requires effective transversal governance solutions. AI has the potential to play a significant role in transnational STI ecosystems by enabling new forms of collaboration, accelerating the pace of discovery and innovation, and enhancing the efficiency and effectiveness of R&D processes, to address interconnected economic, social, and environmental challenges.
Adapting STI ecosystems to benefit from AI deployments presents not only opportunities but also challenges and risks that are dependent on a range of factors including AI enablers, existing structural inequities, the extent of AI adoption, and the level of investment in AI R&D. Quality machine-readable data plays a critical role in AI governance. Thus, data governance is a critical component of AI governance, as the quality and integrity of the data used can have significant implications for the outcomes of the AI system. Effective data governance is crucial to fostering innovation by enabling the development and implementation of AI technologies that are ethical, responsible, and effective in reaching STI policy goals, especially given the breakneck speed with which AI advances such as generative AI are taking place.
Diverging AI impacts and contrasting regional realities
The countries most prepared to reap the benefits from the datafication of socioeconomic activity and technological progress associated with AI are those that are already equipped with the critical digital infrastructure and enabling economic factor endowments associated with higher internet penetration. While the potential risks of AI can be similarly experienced across the world, the Global South is potentially more vulnerable to the estimated harms associated with AI due to transversal systemic constraints, including historical legacies of structural marginalisation.
Countries with the highest number of innovators, shareholders, and investors who provide the intellectual and physical capital to power AI systems will typically be the biggest beneficiaries of AI. This will widen the disproportionate wealth disparity between countries that rely on capital and those that rely on labour and natural resources, including many low and middle-income countries . As evidenced by previous industrial revolutions, the benefits of widespread technological disruptions are distributed unevenly.
2. The G20’s Role
Discussions on responsible AI global governance has become a regular part of the G20 agenda. Several initiatives and working groups have been established where discussions typically focus on three main areas: ethical considerations, economic implications, and regulatory frameworks. While the G20 AI Principles provide a framework for countries and organisations to develop and deploy AI in a way that is beneficial and addresses concerns related to ethics, privacy, and security, there is limited consideration of the distributional aspects and existing multidimensional power dynamics that shape global AI governance.
The G20 has a history of success in creating consensus and coordinating action-oriented inclusive frameworks that support the developmental agenda and leverage the potential of international cooperation despite tensions and sometimes contrasting stances. The G20 has historically also been called upon to create a coordinating committee for the governance of artificial intelligence and data (CCGAID) that simultaneously institutionalises linkages between relevant actors within the G20 and the broader global responsible AI, data, and STI regime, and amplifies multistakeholder participation of the Global South in the development of global AI governance processes.
Inclusive multistakeholder interdisciplinary debates, particularly for potentially disruptive technologies such as AI are crucial to ensure that systemic consolidation of power and control are addressed.
3. Recommendations to the G20
Create an inclusive framework to promote a decolonial-informed approach
Different regions and cultures may have different interpretations of what constitutes ethical AI, which can lead to disagreements on governance. The identification of areas where cross-cultural agreement on norms, standards, or rules and where alternative interpretations and approaches are acceptable or even desirable is a fundamental difficulty in developing inclusive AI ethics and governance.
Furthermore, given that ‘fairness’ and ‘security’ are contested concepts that can lead to disagreements on AI governance policies, we propose the establishment of a Global AI Knowledge Hub, a centralised platform for sharing best practices, research findings, and policy recommendations on AI governance. Such a platform would address ethical issues and involve experts and citizens in the transformation of technical and social considerations into governance mechanisms, thereby benefitting countries of both the Global North and Global South.
As part of a mission-oriented multistakeholder committee, the CCGAID can promote the G20 AI Principles by initiating an inclusive framework that ensures the involvement of multistakeholder representatives from the Global South. It is important that the inclusive framework not simply be exclusive to current G20 members, and that various players from the Global South must be present and heard during both the creation of the CCGAID and the subsequent drafting of technical standards, regulations, and implementation strategies. The inclusive framework is suitable for meta-governance mechanisms for reformed multilateralism. It can also be used to build on existing initiatives such as data-free flows with trust (DFFT),just data value creation (JDVC), and responsible AI.
Given the increasing significance of regional coordination, regional political and economic organisations from the Global South, such as the African Union Commission , Association of Southeast Asian Nations, and the Southern Common Market (better known as Mercosur) should also be included as part of the G20 CCGAID. The CCGAID inclusive framework must be grounded in an inter-vertical approach, which would provide a forum for multilateral policy formulation and mechanisms for collaborative capacity-building, those that analyse, synthesise, and build upon links between varied experiences in different verticals. For example, lessons learned from the successful DFFT, JDVC, and responsible AI of data ecosystems with more AI maturity would inspire solutions to challenges in data-transfers and AI risks in another field. Such reflections would also allow multilateral institutions to adopt reflexivity to identify impending global and societal needs, and also tailor institutional objectives that support existing initiatives.
A decolonial-informed approach (DIA) to responsible AI governance can help address power imbalances, encourage capacity building to legitimise international cooperation, and ensure that the expertise, experience, and perspectives of Global South stakeholders are considered.
To support a DIA, the CCGAID must establish a dedicated Global South Working Group (GSWG) that includes multistakeholder representatives from the Global South. This working group would ensure the inclusion of diverse perspectives in shaping respobsible AI governance frameworks, facilitate capacity building and collaborative learning programmes, provide technical assistance, fund research partnerships, offer scholarships for Global South researchers and policymakers, and facilitate the exchange of knowledge and best practices among participating countries.
The GSWG should also ensure that the different stakeholder groups are sufficiently equipped to navigate the complex landscape of data, digitalisation, responsible AI, and STI and are involved in initiatives to foster international collaboration on crosscutting data, digitalisation, RAI, and STI challenges.
Coordinate transversal policies and agile regulatory frameworks
The CCGAID should establish mechanisms to coordinate policies across different sectors and domains to ensure coherence in addressing crosscutting challenges related to responsible AI, data, digitalisation, and STI governance. The CCGAID should be motivated by financing and capacity-building mechanisms for the implementation of agreed technical standards and regulations that would assist in improving the Global South’s digital public infrastructure, as well as by strategies to guide Global South policymakers in the formulation of domestic responsible AI policy frameworks compliant with global data governance requirements, to foster their STI ecosystems. Regulatory frameworks should also be flexible enough to balance AI innovations with risk management recommendations on AI governance and other mega-trends such as climate change, demographic shifts, urbanisation, digital technologies, and inequalities.
Crosscutting challenges arise due to regulatory fragmentation and lack of interoperability in data ecosystems. Global policies should promote data integration, standardisation, and secure sharing mechanisms to facilitate seamless collaboration and strengthening of multi-level data ecosystems. Transversal governance should address issues of bias, privacy, accountability, ethics, and transparency in AI systems. For developing nations, it is important to co-create conceptual and normative global AI frameworks that align with their specific requirements. By adopting a holistic and collaborative approach, policymakers can tackle the complexity of crosscutting governance domains, promote responsible innovation, and ensure that the benefits of data-driven technologies are harnessed for sustainable and inclusive digital development, suited to various innovation ecosystems.
Collaborate with other multilateral data and AI governance initiatives to prevent duplication of efforts
Several supranational initiatives have been launched to promote responsible AI governance, development, and deployment that support the Sustainable Development Goals. Given the rapid pace of technological advancements, the CCGAID must collaborate with other multilateral organisations to pool resources and leverage investments in capacity-building programmes to enhance stakeholder skills and knowledge in order for them to understand and navigate the complex landscape of data, digitalisation, AI, and STI as well as develop and implement agile regulatory frameworks that keep up with technological advancements while safeguarding the public interest, privacy, and security.
However, there needs to be more emphasis on the global distributional issues associated with AI, which requires a comprehensive approach that involves multiple stakeholders and careful consideration of the potential impacts of AI and the co-creation of policies, regulations, and deadline-oriented implementable interventions, which can mitigate adverse effects, while promoting positive outcomes for all of society.
International organisations collaborating and building on existing multilateral initiatives can identify global opportunities and guide future investments. They can strengthen oversight, facilitate research on common challenges, and promote the sharing of best practices to govern generative AI, foundational models, and data to support responsible AI governance. Organisations such as the World Bank, Organization for Economic Cooperation and Development, and the United Nations can work together with the G20 to establish clear objectives and identify areas of overlap or complementarity in global data, digitalisation, and responsible AI initiatives.
This collaboration must include multilateral developmental partnerships with developed nations that have already implemented AI governance frameworks, which can provide valuable cross-sectoral lessons and knowledge exchange. Regular consultations, joint research projects, and the sharing of best practices and expertise can facilitate this collaboration. Furthermore, as part of the inclusive framework, the CCGAID should facilitate initiatives with other multilateral institutions on technology transfers to bridge the data and digital divide and ensure equitable access to AI capabilities, particularly in the Global South.
 The G20 Financial Stability Board (FSB) and The OECD/G20 Base Erosion and Profit Shifting (BEPS) initiative
 Walter Mattli, “The politics and economics of international institutional standards setting: an introduction,” Journal of European Public Policy 8, no. 3, 2001, 328-344.
 Yuzhuo Cai and Marcelo Amaral, “The Triple Helix Model and the Future of Innovation: A Reflection on the Triple Helix Research Agenda,” Triple Helix 8, no. 2. 2021. 217-229.
 Kutoma Wakunuma, George Ogoh, Damian Eke, and Simi Akintoye, “Responsible AI, SDGs, and AI Governance in Africa,” 2022 IST-Africa Conference (IST-Africa), 2022.
 Chux U Daniels, Olga Ustyuzhantseva, and Wei Yao, “Innovation for inclusive development, public policy support and triple helix: perspectives from BRICS,” African Journal of Science, Technology, Innovation and Development 9, no. 5.2017,513-527.
 Raphael Kaplinsky and Erika Kraemer-Mbula, “Innovation and Uneven Development: The Challenge for Low- and Middle-Income Economies,” Research Policy 51, no. 2.2022.
 Daron Acemoglu and Simon Johnson, Power and Progress: Our Thousand-Year Struggle Over Technology and Prosperity (New York: PublicAffairs, 2023)
 Yadong Luo, “Illusions of techno-nationalism,” Journal of International Business Studies 53, 2022, 550-567
 Jing Cheng and Jinghan Zeng, “Shaping AI’s Future? China in Global AI Governance,” Journal of Contemporary China, 2022.
 Ren Bin Lee ,Dixon, “A principled governance for emerging AI regimes: lessons from China, the European Union, and the United States”, AI Ethics, 2022.
 Charlotte Siegmann and Markus Anderljung, “The Brussels Effect and Artificial Intelligence: How EU Regulation Will Impact the Global AI Market” Centre for the Governance of AI, 2022.
 LewinSchmitt, “Mapping global AI governance: a nascent regime in a fragmented landscape. AI Ethics 2, 2022, 303–314
 Dani Rodrik, “When Ideas and Technologies Cause More Harm than Good,” Project Syndicate, February 9, 2023.
 Srinivasan Thirukodikaval Nilakanta and Quibria Muhammed, “Technological Progress, Factor Endowments and Structural Change: A Note,” The Bangladesh Development Studies 32, no. 4, 2009, 95–105
 United Nations Conference on Trade and Development, “Structural Transformation, Industry 4.0 and inequality: Science, Technology and Innovation Policy Challenges,” UNCTAD, 2019.
 Chinmayi Arun, “AI and the Global South: Designing for Other Worlds,” In Markus D. Dubber, Frank Pasquale, and Sunit Das (eds.), The Oxford Handbook of Ethics of AI (Oxford: Oxford University Press, 2019).
 Anton Korinek, Martin Schindler, and Joseph Stiglitz. “Technological Progress, Artificial Intelligence, and Inclusive Growth,” IMF Working Papers 2021/166, 2021.
 UNCTAD, “Technology and Innovation Report”
 Thorsten Jelinek, Wendell Wallach, and Danil Kerimi, “Policy Brief: The Creation of a G20 Coordinating Committee for The Governance Of Artificial Intelligence,” AI and Ethics, 2021,141-150.
 Mariana Mazzucato, “Catch-up and Mission-oriented Innovation,” in Arkebe Oqubay and Kenichi Ohno (eds), How Nations Learn: Technological Learning, Industrial Policy, and Catch-up (Oxford: Oxford University Press, 2019)
 Fumiko Kudo, Ryosuke Sakaki, and Jonathan Soble, “Every Country Has Its Own Digital Laws. How Can We Get Data Flowing Freely Between Them?,” WeForum, May 20, 2022.
 Shamira Ahmed, “Rethinking Data Governance for Just Public Data Value Creation and Responsible Artificial Intelligence in Africa,” Medium, May 5, 2023.
 United Nations Environment Programme, “Report for the UN 75th Anniversary: Shaping the Trends of Our Time,” UNEP, September 17, 2020.
 Ekkehard Ernst, Rossana Merola, and Daniel Samaan, “The economics of artificial intelligence: Implications for the future of work,” ILO Future Of Work Research Paper Series, 2018.