AI's promise meets the reality of work and productivity

Artificial intelligence is reshaping the economy, but translating its potential into broad-based productivity gains will take time, while governments and businesses face growing pressure to manage its impact on workers and ensure its benefits are widely shared, participants at the 2026 IMF-World Bank annual meetings in Bangkok said this week.

Although AI has transformed consumer-facing services, its adoption in high-stakes business operations remains difficult. Many corporate AI projects have fallen short because they add tools to existing processes rather than fundamentally redesigning how work gets done. Supply-chain constraints and social resistance also threaten to slow the technology's expansion.

Kristalina Georgieva, managing director of the International Monetary Fund, said AI could challenge a historical pattern in which technological revolutions have generally created more jobs than they destroyed.

"This time, even the new jobs may be taken over by AI," she warned, pointing to the uncertainty surrounding the technology's long-term impact on employment.

Although current data does not yet show widespread job losses, Georgieva said, changes in labor demand could expose deeper social tensions. Rising incomes among highly skilled workers may fuel demand for lower-paid services, even as younger workers struggle to find jobs that match their career expectations.

Previous technological transitions were accompanied by promises of retraining that often failed to materialize, contributing to entrenched social divisions and the rise of populism. Avoiding a repeat will require coordinated action, she said.

"Companies need to retrain their workers and change their business models," Georgieva said, calling on governments to communicate risks transparently and consider subsidies to help businesses absorb the costs of transition.

The broader challenge is that AI's productivity effects remain at an early stage. Improvements so far have been incremental rather than transformative, raising the risk that companies could abandon investments if returns fail to meet short-term expectations.

Jack Clark, co-founder and head of public benefit at AI research company Anthropic, argued that productivity should not be measured solely by how quickly output increases. AI could also improve decision-making, while automating routine information processing and enabling physical applications such as robotics could deliver gains across a much wider range of jobs.

Clark identified four prerequisites for inclusive AI adoption: digital infrastructure and reliable electricity, a skilled workforce, mechanisms for spreading the technology across industries, and effective risk governance. Infrastructure gaps in parts of Africa remain a significant obstacle to distributing AI's benefits more evenly.

"AI is not going to benefit everyone automatically," Clark said, stressing the need to address these underlying conditions.

The speakers also called for common standards to build trust and facilitate trade. Reciprocal arrangements modeled on international toy-safety standards could offer a precedent for AI governance. Modular system architectures, meanwhile, could help small and medium-sized businesses participate in the AI economy rather than allow gains to concentrate among a handful of technology giants.

With AI advancing faster than traditional regulation can keep pace, industry self-governance and public-private cooperation are becoming increasingly important. Clark compared the challenge to Formula One racing.

"We cannot ban racing because we are afraid of accidents, but we cannot race without guardrails either," he said, emphasizing the importance of technical standards, testing requirements, and transparency.

Global coordination remains difficult, however, as major AI-producing countries — including the United States, China, and India — hold differing positions on governance. Without effective safeguards, Clark warned, the risks associated with increasingly powerful AI systems could continue to accumulate.