The recent explosion in automation enabled by artificial intelligence, often dubbed the fourth industrial revolution, is fuelling expectations of a future of ‘phenomenal wealth ^[1]’, including an economy in which the price of labour will be close to zero. AI is expected to contribute US$15.7 trillion ^[2] to global GDP by 2030—more than the current economic output of India and China combined.

But will this wealth be created for everyone? Not necessarily.

The International Monetary Fund predicts ^[3] that AI-enabled automation will severely widen the economic gap between developing and advanced economies. Even within countries, regional disparities in wealth are anticipated to soar as the use of AI technologies remains concentrated in select urban centres ^[4]. If AI is indeed on track to create a more unequal world, how can global governance institutions address the potential crisis?

The global AI governance regime is taking shape under the influence of two predominant forces: the Organisation for Economic Co-operation and Development (OECD) and the UN Educational, Scientific and Cultural Organisation (UNESCO). Both aspire to play a defining role in shaping international standards and best practices for responsible AI.

The OECD released its recommendation on artificial intelligence ^[5] in May 2019, while UNESCO released its recommendation on the ethics of artificial intelligence ^[6] in November 2021. The OECD and UNESCO recommendations are attempts to provide a comprehensive collection of normative prescriptions to establish a global standard for responsible AI. Both intergovernmental agencies are now developing strategies to facilitate implementation, individually as well as through cooperative engagement at the Global Partnership on Artificial Intelligence ^[7] (GPAI).

Important but not well known, the GPAI was launched in 2020 as a new thematic, multilateral, multi-stakeholder forum with the stated mission to ‘bridge the gap between theory and practice’ on responsible AI. With AI superpowers such as the United States and Canada included in its membership, the OECD serving as its secretariat ^[8] and UNESCO as observer ^[9], GPAI is on the fast track to become a formidable global reference point for responsible AI. But its aspiration to secure global cooperation for trustworthy adoption of AI remains unlikely in the face of a growing global divide ^[10] in AI between the northern and southern hemispheres.

India, as a fast-emerging AI superpower and the incoming council chair ^[11] of GPAI, must push for equitable north–south resource-sharing to bridge the divide. This will be critical to ensure consensus-based cooperation on responsible AI and its global adoption.

The absence or underrepresentation of the global south in forums and initiatives on responsible AI is primarily a symptom of its lagging AI adoption rate (with the exception of China and India ^[12]). This compromises the developing world’s ability to secure a seat at the AI table, despite its significant contribution ^[13] to the global AI industry.

Only four out of the 29 GPAI member countries are from the global south ^[14]: Argentina, Brazil, India and Senegal. Representation in the GPAI’s expert working groups (whose members need not be nationals of GPAI members ^[15]) also remains minuscule. Even at other global forums on AI ethics and governance, the global south, with the exception of India and China, has remained largely absent or underrepresented ^[16]. For example, only five of the 179 initiatives ^[17] on AI ethics and governance have their origin in the global south.

Much of the global south continues to score low ^[18] in government AI readiness, and none ^[19] of the low- and middle-income countries boast an AI strategy yet. If one leaves out India and China, none ^[20] of the world’s top 100 AI companies are headquartered in the global south. The region contributes less than 5% ^[21] to the global AI talent pool.

Fewer than 20 ^[21] of the world’s 500 fastest (non-distributed) supercomputers on which AI algorithms are developed and tested are collectively hosted in the global south. Most low- and middle-income countries remain deprived of basic digital infrastructure and connectivity, let alone supercomputers. Most countries in the region also continue to lack critical resource capacity to develop adequate regulatory protections against predatory data and AI practices ^[22] by leading technology powers.

The combination of these factors means the global south not only finds itself locked out of the economic growth stemming from AI adoption but also finds itself in a precarious situation. Its role in the new AI-driven global economic order might get relegated to that of a supplier of unscrupulously sourced ^[23] commercial datasets, cheap digital labour ^[24], dangerously mined ^[25] rare-earth minerals (like cobalt and lithium) and dual-use AI testbeds ^[23]. Because of this ‘AI colonialism ^[26]’, developing countries that don’t have access to these technologies may be left behind, creating a new global divide that reinforces existing power structures and perpetuates economic inequalities.

India, as the incoming GPAI council chair, should work with likeminded partners such as Australia, the US, Japan and France to address the growing north–south AI divide. A coordinated approach would promote inclusive growth in the Indo-Pacific region, which is home to many low- and middle-income countries that could gain from the economic benefits of AI adoption but that also need resources to combat algorithm-driven exploitation.

GPAI’s nascence gives it room to experiment and steer global policy reform to level the playing field for all countries pursuing responsible AI adoption. With its commitment to amplify the voice ^[27] of the global south, India must leverage its GPAI leadership to steer concrete interventions. This will be critical to facilitating expansion ^[28] of the global south’s membership of and participation at GPAI.

India should look at scaling the initiative to connect small and medium-sized enterprises to providers of AI solutions through the AI4SME portal ^[29], a crucial and promising effort by GPAI to help strengthen and scale business adoption of AI in member countries. The AI4SME network should be expanded to include individual and institutional investors, researchers and professionals, and startup mentors—all of whom continue to be in short supply in the global south.

Initiatives such as the Partnership for Global Infrastructure and Investment ^[30] could play a critical role in financing the infrastructure requirements for AI development and use in the global south. GPAI should commission research on how such cooperative north–south AI financing arrangements should be structured to tackle the risks of new global digital divides. Research should be expanded to include the need for reforms in international trade, investment and intellectual property regimes to make sure that the pursuit of global innovation and prosperity doesn’t come at the cost of equity.

GPAI’s working group on the future of work should support the design and implementation of data- and AI-centric teacher and faculty development programs at schools and universities in the global south, so that its industrial workforce can become future-ready, globally competitive and less vulnerable to job loss and exploitation. This effort should be complimented by workshops for public administrators in AI-deprived countries to help them develop action strategies for AI adoption. The workshop modules should be modelled on GPAI’s ​and UNESCO’s existing ^[31] frameworks ^[32] for educating policymakers on AI adoption.

Global leaders and policymakers, including Indo-Pacific powers such as India and Australia, must recognise and take concerted measures to address the fundamental inequities that continue to persist in how AI is built, deployed, used and traded globally. Only then will the aspiration for inclusive growth and global adoption of responsible AI be taken seriously.