Commentary: The world’s argument about artificial intelligence has been framed too narrowly. Governments debate safety rules, companies race to build data centres, and investors ask whether the boom is sustainable. The more consequential question is who will benefit when the machines become more capable—and who will pay when they do not.
Recent warnings from the International Monetary Fund should sharpen that debate. IMF Managing Director Kristalina Georgieva has described AI as both a positive demand shock and a source of inflation, while warning that a reversal in AI investment could threaten growth and financial stability. The fund has also cautioned that rapid technological change may widen the gulf between advanced and developing economies. Reuters reported those warnings in October.
Our view is straightforward: governments should stop treating AI primarily as either a private-sector opportunity or a regulatory hazard. It is becoming public infrastructure. That means the gains from its productivity should be broadly shared, and the costs of disruption should not be quietly transferred to workers, poorer countries and future taxpayers.
The case for speed
The arguments for rapid deployment are real. AI investment is supporting demand at a moment when the global economy faces weak growth, high debt and costly energy shocks. It could improve medical research, education, public administration and energy systems. Slowing development in the name of perfect certainty could leave societies less competitive and deny people useful tools.
Businesses also have a legitimate concern that fragmented national rules will raise costs and entrench the largest technology firms. Common standards, predictable liability rules and access to skilled workers could encourage responsible innovation rather than drive it into less accountable jurisdictions.
Those considerations deserve more than caricature. A society that blocks every new technology will not protect ordinary people; it will simply ensure that decisions are made elsewhere.
Why the market alone is insufficient
But speed is not a social policy. AI’s benefits will flow first to firms that control computing capacity, data, distribution and capital. Its disruptions will be felt most immediately by people whose tasks can be automated, by small companies unable to compete for infrastructure, and by countries that lack the money to train workers or build domestic systems.
The IMF has warned that AI and other shocks threaten to deepen inequality between advanced and developing economies. At the same time, debt-service costs for low-income countries have risen sharply, limiting their ability to invest in education and digital capacity. Reuters’ account of the IMF and World Bank meetings described developing countries leaving with unresolved questions about debt, war and a possible AI bubble.
This is not an argument against innovation. It is an argument against pretending that innovation automatically distributes itself fairly. Without deliberate choices, AI could increase productivity while weakening bargaining power, concentrate wealth and leave poorer nations dependent on systems they neither control nor fully understand.
What a fair bargain would require
First, governments should require serious transition planning from companies benefiting from public incentives, including credible investment in worker training and independent evaluation of job impacts. “Reskilling” cannot remain a slogan attached to short online courses; it must include paid learning, portable credentials and stronger social protection.
Second, public money for computing infrastructure and research should come with public-interest conditions. Governments should support open scientific work, secure access for universities and smaller firms, and interoperable systems that prevent a handful of companies from becoming unavoidable gatekeepers.
Third, international institutions should treat digital capacity as development policy. Wealthier countries can help finance connectivity, education and locally governed AI tools instead of asking poorer states only to adopt standards written elsewhere.
Finally, regulators should be willing to challenge market power. The objective is not to punish scale, but to ensure that scale does not become a licence to dictate prices, suppress competition or evade accountability.
There is a counterargument: redistribution and regulation could reduce investment precisely when economies need growth. That risk is genuine. Yet refusing to set conditions creates its own danger. If the next downturn follows an overheated AI investment cycle, as international officials fear, societies with concentrated gains and weak safety nets will absorb the heaviest losses.
The AI boom should not be stopped. It should be governed as a shared economic transformation. Growth matters, but legitimacy matters too. If governments want citizens to trust artificial intelligence, they must make a visible bargain: those who gain most from the technology must help equip everyone else to live with it.