Global Disparities in AI Development

AI hold immense promise for reducing global inequalities by enhancing productivity, diversifying economies and closing skills gaps in developing nations. Success stories like India’s “Saagu Baagu” initiative demonstrate AI’s potential to drive meaningful progress and change. However, the dominance of HICs in AI development poses risks of dependency and inequality, with poorer nations potentially burdened by high costs and limited control. To truly bridge the global development gap, we must ensure that AI’s benefits are shared equitably. Collaborative efforts in education, infrastructure and regulation are essential to harness AI’s potential whilst avoiding the pitfalls of technological dependency. The challenge ahead is to guide AI as a tool for global inclusion, not division.

The development of AI models mostly occurs in HICs due to the very high level of skills and expensive capital required to create complex technology. In fact, according to the Stanford University AI Index Report the only non-HIC countries that produce AI models are China and Egypt. Because of this fact, many poorer nations, who seek to use AI to fuel development, may become dependent and reliant on richer nations to provide this technological product. This may be particularly problematic if richer nations view these developing nations as a threat to their global competitiveness. Here, they may enforce tougher trade regulations which constrict the supply of AI products to less wealthy countries, dampening growth in the developing world. Another issue associated with AI is that the sector is heavily privatised, driven by a lack of regulation and low barriers to entry. To analyse this, we can look at the number of AI models released per year by various sector, as well as the complexity of these models (measured using the number of parameters – a higher value indicates a more complex model and vice versa) using data from the Stanford University AI Index Report. Especially in recent times, the development of AI models has been dominated by industry and industry partnerships led by private firms.

Additionally, these industry-led models generally have more parameters and hence are more complex, resulting in them having far greater capabilities than those produced by academia and governments.

Because industry produces more models which are of higher quality, developing countries are likely to purchase these better AI packages. As they buy these AI products from private firms (predominantly in HICs) who seek to maximise profit, significant costs will be incurred on poorer nations. Given that these countries struggle to fund basic goods and services like healthcare, education and infrastructure, it is questionable whether developing nations can afford these AI strategies in the first place.

The Role of AI in Driving Economic Diversification

One major challenge facing developing nations is limited economic and export diversity. This can be measured using the Economic Diversification Index, a metric developed by the Mohammed bin Rashid School of Government. A higher score indicates greater diversification and vice versa. The data suggests less developed countries typically have a less diversified economic structure, with the largest gap being between UMIC and HIC Countries. This stems from the fact that in developing countries, the labour force often lacks the skills or training to move into other sectors from primary and secondary industries. For example, UNESCO in 2023 found the GER (Gross Enrolment Ratio) for tertiary education in LICs to be just 9% – by contrast, in HICs it was 80%. As a result, growth is slow, incomes remain low, and the economy becomes more prone to shocks in the global market. AI could play a key role in closing this skills deficit. The mechanism of AI means it can “learn” from sources like the internet and higher education institutions in richer countries, and transfer the relevant information to less wealthy countries through question-and-answer prompts, or use this information in autonomous decision-making processes targeted at improving efficiency in emerging sectors. As such, there is a flow of skills and data from the developed to developing world, helping diversify lower and middle-income countries as this skills gap is closed. This may also be of benefit to HICs in the Middle East like Qatar, Kuwait, the United Arab Emirates, Bahrain and Oman, which all have relatively undiversified economies due to their dependence on oil exports. Here, AI could provide a low-cost solution to finding and providing skills to allow their economies to branch out in the face of dwindling oil reserves and environmentalist pressures.

Another obstacle hindering growth in developing countries is the slack in productivity levels. According to the ILO (International Labour Organisation), in 2023 LICs outputted $2.91 per hour worked, LMICs outputted $8.47 and UMICs outputted $17.27. In comparison, HICs outputted a staggering $57.99 per hour worked. AI could play a key role in closing this gap. Its effect on productivity can be analysed by plotting output per hour worked against the number of AI models in 3 areas – the US, EU (including the UK) and China – using data from both the ILO and Stanford University’s AI Index Report.

The US, EU and China operate very differently, with vast disparities in the levels of protectionism and interventionist measures, yet all see a positive correlation between these two variables, experiencing a rise in productivity due to the growth in AI. It would be remiss to suggest this has been the sole cause of productivity rises, however AI growth has certainly been one of many contributing factors increasing the productive capacity of these nations. Hence, if developing nations adopt AI technologies, there is great potential to reap the same benefits as these three regions, increasing productivity and reducing the gap between them and the developed world.

A Case Study: India’s “Saagu Baagu” Agricultural Initiative

A key case study to look at here is the implementation of AI in agriculture. Specifically in developing nations, the primary sector has suffered from an array of climate-related and economic challenges. To combat this, the World Economic Forum introduced the AI for Agricultural Innovation (AI4AI) scheme, its most successful application being the “Saagu Baagu” initiative in Telangana, India. The government, through partnerships with firms including Digital Green and Glific, developed an AI chatbot that provided farmers with suggestions on how to maximise their crop health; established soil testing centres equipped with machine learning, which offered farmers insight into soil health and fertiliser recommendations; and introduced a computer vision system which assessed the quality of crops, determining defects and identifying attributes like colour, shape and size. The scheme yielded remarkable results, with farmers seeing a 21% increase in crop yield per acre. Pesticide use reduced by 9% and fertiliser use fell by 5%, significantly reducing costs. Due to quality improvements, unit prices for crops increased by 8%, improving farmer incomes by around $800, almost doubling their earnings. This example demonstrates how “agritech” and AI can greatly improve output in the primary sector within developing nations, and given that – according to the World Bank – in LICs agriculture makes up 25.58% of GDP compared to just 1.28% in HICs, this could prove to be a powerful tool in bridging the global development gap.

Is AI bridging or widening disparities in global development?

The swift evolution of artificial intelligence is reshaping both production and economic landscapes, bringing profound changes to economies around the globe and redefining international trade dynamics. As we navigate this contemporary industrial revolution, it’s crucial to explore AI’s role in economic development, especially its promise to spur growth in developing nations. Indeed, there is a growing belief that AI could play a pivotal role in addressing one of the most persistent challenges in economic history: bridging the global development gap.

Figures from the Stanford AI Index Report, unless stated otherwise