AI risks may exceed opportunities without safeguards and ethical code: IMF chief

The risks of artificial intelligence (AI) may exceed its opportunities if no adequate safeguards and ethical framework are put in place, International Monetary Fund (IMF) chief Kristalina Georgieva said on Friday, backing a blend of industry self-regulation and input from key governments.

The nascent technology was a dominant topic at this week’s annual meetings of the IMF and World Bank in Bangkok, where the IMF on Tuesday flagged significant growth and financial stability risks from a potential end of the AI investment boom, high debt levels ‌and a prolonged war in the Middle East.

During a panel discussion that included Anthropic co-founder Jack Clark, Georgieva said AI was already strengthening the global economy but could also deliver negative shocks if it failed to deliver major productivity gains as quickly as hoped.

“If there are no safeguards, then the risks may exceed the opportunities. And above all, we do need an ethical foundation, and I’m not sure we are there as a society,” she said.

Certain preconditions needed to be put in place for people everywhere to benefit from the new and rapidly evolving technology, including access to digital infrastructure, access to electricity and adequate regulation, Georgieva said.

The bigger question now was how to create guardrails for a technology that was evolving so rapidly, she said, likening AI developments to Formula One race cars careening around with no barriers to protect the audience.

Industry self-regulation was appropriate, given that governments were not well-placed to regulate or oversee the constantly changing sector, but should include engagement with society. The countries with the biggest AI sectors, including the US, China and India, along with countries in the AI supply chain, should be involved, she told the panel.

Institutions such as the IMF should also be involved in their areas of competence, such as productivity growth, labour markets and financial stability risks, Georgieva said.

“I think it’s Formula One at night, without maybe brakes in the cars,” quipped Anthropic’s Clark, picking up her analogy. He argued that setting standards was possible and was already starting through early policy frameworks in the EU and voluntary commitments in the US.

Clark said his company regularly tested its systems, published details on the tests and shared the information, but he did not believe it should be completely voluntary.

Asked if he would favour regulation from the countries that produced AI, Clark said: “Yes.