Georgieva’s metaphor: a race without sufficient protections
Kristalina Georgieva, Managing Director of the International Monetary Fund, called for introducing “barriers” to contain the risks of the rapid expansion of artificial intelligence. During a panel in Bangkok held as part of the IMF and World Bank meetings, she compared the technology to a Formula 1 car that could leave the track and hit people. The message, reported by the Associated Press, at the same time acknowledges the possibility that AI could bring significant benefits.
Jack Clark, co-founder and Head of Policy at Anthropic, took part in the discussion. Clark called for broader public involvement in decisions and argued that companies alone should not set the governance framework. His remarks reflect the position of an executive at a lab developing advanced models: they are a contribution to the debate, not an independent assessment of risk.
The economic question is distribution, not just growth
According to Georgieva, AI’s impact on work remains uncertain. She described different categories: transformed or new jobs, repetitive tasks that could shrink, and services that expand thanks to investment. An overall increase in economic activity does not guarantee that the gains reach the same people who lose opportunities or bargaining power. Bank of Canada Governor Tiff Macklem cited the precedent of trade liberalization: aggregate benefits can coexist with losses concentrated among specific workers and territories.
In its studies, the IMF distinguishes between jobs’ exposure to automation and to productivity gains. The actual extent will depend on how companies redesign processes, on available training and on support policies. For this reason, the issue is not exhausted by asking whether a model can perform a task: what matters is who decides how it is introduced and who benefits from the additional productivity.
Infrastructure and cybersecurity matter as much as models
Georgieva noted that distributing AI’s benefits requires electricity and access to digital networks. Without reliable energy, connectivity and skills, adoption risks concentrating in countries and companies already better equipped. The panel also highlighted cybersecurity, because systems with access to data and tools can amplify vulnerabilities as well as create efficiencies.
The “barriers” called for are not yet a single regulatory plan. They can include testing, transparency obligations, capability checks, liability for high-impact uses and protections for workers. The common point is that measures must be proportionate to risks and applicable even when technology changes rapidly.
The IMF’s warning brings AI into macroeconomic discussion: growth, employment, inequality and stability. Estimates remain uncertain, but decisions on infrastructure and rules are already under way. The challenge will be to distribute the benefits without leaving companies and citizens alone in the face of transition costs.
