Janhavi Pathak | 12 September 2024

Summary

• Skewed access to Artificial Intelligence (AI) resources and development capabilities will aggravate global socio-economic inequalities. Resource-rich countries will drive Research and Development (R&D) predicating future AI progress, and accruing significant economic gains in the long run. 

• Formulating sovereign AI policies, building infrastructural capacity and localising foundational models will become a priority for resource-limited countries and achieved through collaborative and multistakeholder public-private initiatives. 

• Due to brain-drain, sustaining and retaining an AI-skilled and reskilled workforce will be difficult for resource-limited countries. Stringent environmental and regulatory laws in the Global North may gradually shift AI data centres to the Global South, which can incentivise the development of sovereign AI strategies in the region, while adversely impacting decarbonisation goals.

AI Ecosystems in Global North vs Global South

Resource-rich countries of the Global North host most of the cutting-edge AI technology, propelled by sustained long-term investment in R&D, adequate infrastructure capabilities and workforce capacity. A handful of big tech companies, often based in high-income countries, drive these undertakings and accrue concomitant profits. In parallel, resource-limited countries in the Global South grapple with the lack of necessary infrastructure and human resource capacity, including nationwide network connectivity and an AI-trained workforce. This inevitably has impeded the optimal incorporation of AI in the economy despite the opportunities made available by open-source AI platforms.

Consequently, countries in the Global South rely heavily on the big tech giants to provide access to emerging AI technology through state collaborations and private sector joint ventures. A direct result of this is reflected in the inability of some foundational Large Language Models (LLMs) created in high-income countries to solve the unique and industry-specific problems facing low to medium-income states. To address this, several states in the Global South, such as India, China and Brazil have drafted and implemented national AI legislation whilst focusing on building localised AI models and resolving their challenges. Politically, localised models and indigenous data centres would empower states to claim sovereign ownership over their data and intelligence. 

Challenges to AI Democratisation 

AI inequity runs the risk of perpetuating and aggravating global socio-economic differences between the Global North and South. An increase in labour productivity through innovation and risk-taking behaviour will eventually be translated into improved economic performance for countries undertaking strategic investment in the sector. Despite pledges of climate-focused initiatives while establishing data centres in the Global South by tech giants and national governments, countries in these regions bear a disproportionate environmental burden due to an absence of comprehensive regulatory and environmental oversight to ensure compliance.

Skewed AI access to emerging tech will stymie the economic performance and competitive edge of countries lagging due to an unequal playing field. Rising calls for AI democratisation question the increasing centralisation, seeking more equal distribution of generated profits, technology autonomy and privacy for the Global South. As governments in the region legislate sovereign AI policies to build a strategic roadmap for future developments and AI integration, a compelling need for circumstantial calibration is required to harness individual strengths and overcome weaknesses. 

Ownership of sovereign data and intelligence would be crucial in offsetting the growing power imbalance between the ‘AI-leading’ and ‘AI-lagging’ states. Democratisation needs to transcend open-source modelling to include conscious multi-stakeholder efforts toward regional integration and profit distribution. In other words, access to functions should go hand-in-hand with benefits availed for people living in different parts of the world. The well-being/cost ratio for developing and deploying AI needs to be fair.

Forecasts

• Medium-term

  • The development of cutting-edge AI technology will most likely remain concentrated in resource-rich countries, predominantly in the Global North. 

  • An expansion of sovereign AI strategies will likely crystallise in the Global South, as private-public partnerships aim to localise emerging AI technologies through sustainable investment in infrastructure and capacity-building measures. 

  • Brain-drain, lack of adequate hardware and software capabilities, and socio-economic and political instability will most likely adversely impact AI progress in the Global South.

• Long-term

  • Unequal access to resources and AI-building capabilities will most likely widen the economic chasm between the Global South and North countries. 

  • Collaborative initiatives to level the AI playing field will likely emerge at the international level, and will most likely witness a greater role of leading private AI companies. 

  • AI sovereignty will likely become a crucial part of ensuring national sovereignty, as countries strive for self-sufficency in production and consumption capabilities.