Economists urge action to prevent ‘AI poverty traps’

Artificial intelligence could deepen inequality and create ‘AI-poverty traps’ in developing nations, write economists Dr Asha Sundaram and Dr Dennis Wesselbaum in their paper ‘Economic development reloaded: the AI revolution in developing nations’.

Sundaram, an associate professor at the University of Auckland Business School, and Wesselbaum, an associate professor at the University of Otago, say developing countries lack the necessary infrastructure and skilled labour force to capitalise on AI's potential.

"The downside is that there isn't a lot of capacity in some countries in terms of digital infrastructure, internet, mobile phone penetration," says Sundaram.

"Much of the technology is controlled by firms like Google and OpenAI, raising the risk of over-reliance on foreign tech, potentially stifling local innovation."

Without strategic interventions, Wesselbaum says AI may create an AI-poverty trap: locking developing nations into technological dependence and widening the gap between global economies.

“For developing countries, AI could be a game-changer; boosting productivity, expanding access to essential services, and fostering local innovation – if the right infrastructure and skills are in place.”

Financial support from developed countries and international bodies like the UN could help cover upfront costs through grants, loans and investment incentives, according to the research.

“We also need robust legal and regulatory frameworks to support responsible AI by addressing data privacy, ethics, and transparency concerns,” says Sundaram.

The economists argue that in developing AI policies, the international community must learn from the successes and failures of foreign aid.

"Aid has often failed to spur lasting growth in developing countries,” says Sundaram, “partly because it can create dependency, reducing self-reliance and domestic initiatives."

She highlights a need for policies to mitigate the downsides of AI, both in developed and developing countries.

Such policies could include an international tax regime that would allow countries to capture tax revenue from economic activities driven by AI inside their borders.

Sundaram’s involved in one such project in Ethiopia where artificial intelligence is being harnessed by the government and the country’s largest telecom provider to support small businesses excluded from formal banking due to lack of collateral.

By analysing mobile money transactions and how much these businesses pay and receive, algorithms estimate how much credit can safely be offered, enabling small loans and helping integrate marginalised enterprises into the formal economy.

Artificial intelligence holds the power to transform development trajectories, but without targeted investments and inclusive policies, says Wesselbaum, it risks deepening the digital divide and entrenching global inequality.