The International Monetary Fund (IMF) is warning that artificial intelligence (AI) could boost global economic growth but expose economies to significant risks if governments and technology companies fail to establish adequate safeguards.
IMF Managing Director Kristalina Georgieva estimates that AI adoption could add about half a percentage point to annual global economic growth over the next decade. However, she cautioned that the benefits would depend on how widely businesses and economies adopt the technology.
“The world can’t afford to get AI wrong,” she wrote in a blog on Friday, noting that current gains were concentrated particularly in economies supplying the hardware behind the technology.
Speaking at a seminar in Bangkok, Ms Georgieva compared AI development to a car race, arguing that safeguards were needed to prevent the technology from causing serious harm.
“We are racing, and the question is how do we install barriers, so if a car was to fly out, it doesn’t kill people,” she said.
She acknowledged that governments lacked sufficient knowledge to regulate AI at the pace required, suggesting that industry-led self-regulation was appropriate for now. She identified the United States, China and India as key players in shaping such efforts.
However, Jack Clark, co-founder and head of public benefit at AI company Anthropic, argued that voluntary safeguards would not be enough.
“I think it’s worse than you described. I think it’s Formula One at night without maybe brakes in the cars,” he said, calling for common-sense standards and mandatory testing.
“Testing shouldn’t ultimately be voluntary,” Mr Clark said, adding that decisions about AI safety should not be left solely to technology companies.
Adoption holds the key to economic gains
Despite growing investment in AI, its wider economic benefits remain uncertain as businesses work to integrate the technology into their operations.
Ms Georgieva said the world was seeing incremental improvements rather than transformation at scale, indicating that the expected productivity gains would depend on adoption across industries.
Tiff Macklem, Governor of the Bank of Canada, also highlighted the importance of spreading AI applications across businesses rather than concentrating the benefits among technology developers and hardware suppliers.
Tobias Meyer, Chief Executive of DHL Group, pointed to the growing use of AI agents that draw on company databases to provide specific information and support operational decisions.
For DHL, which handles about 100 million parcels daily, such applications can help track shipments, process large volumes of customs documentation and respond to customer enquiries.
The example illustrates how AI could improve efficiency in complex operations. However, extending similar benefits across the wider economy will require investment in infrastructure, systems and workforce skills.
Digital divide threatens unequal gains
Ms Georgieva warned that unequal access to digital infrastructure could leave lower-income countries behind as AI adoption accelerates.
According to her blog, advanced economies are three times better equipped digitally than low-income countries, highlighting disparities in their capacity to adopt and benefit from the technology.
Reliable electricity, digital connectivity and a trained workforce are among the requirements for businesses and public institutions to use AI effectively.
For developing economies, these constraints could limit productivity gains even as the technology becomes increasingly accessible through smartphones and other digital devices.
Addressing the divide will therefore be important if AI is to support broad-based economic growth rather than concentrate benefits in countries and businesses that already possess advanced digital capabilities.
In her plenary address to the IMF and World Bank Group annual meetings, Ms Georgieva said accessible AI tools could help widen participation in the technology.
“Simple AI, accessible via a smartphone, could prove very democratic,” she said, invoking science fiction writer William Gibson’s observation that the future is already here but is not evenly distributed.
