Nobel economist Daron Acemoglu finds no measurable AI productivity effect in aggregate data — despite years of hype and hundreds of billions in investment.1
His assessment is direct: AI agents handle fragments of work well. They cannot absorb an entire job.1 "There's a huge amount of uncertainty," Acemoglu told MIT Technology Review, pointing to conflicting signals — anecdotes of worsening graduate job markets alongside flat productivity metrics.1 His forecast: AI gives only a small boost to US productivity and will not eliminate human work.2
That assessment lands at an expensive macro moment. Services inflation remains above 3% annually.3 The Iran conflict has added $857 to Americans' average annual gasoline costs in 2026.3 US 30-year Treasury yields have crossed 5%. UK gilts trade at levels last seen in the 1990s. The Federal Reserve faces a leadership vacuum: Chair Powell's term is expiring and economist Miran has resigned.
AI optimists have relied on a coming productivity dividend to argue current inflation is transitory. The logic: efficiency gains would ease cost pressures without tighter policy. Acemoglu's evidence undermines that case. The transformative application layer — the one that converts AI capabilities into economy-wide output gains — has not appeared.1
For workers, augmentation is not the same as elimination. Productivity tools raise output on specific tasks: drafting, coding, summarization. They do not replace the judgment, coordination, and contextual work that defines most roles. That gap between task-level performance and job-level replacement is central to Acemoglu's argument.
Entry-level workers feel the disconnect most acutely. Anecdotes of tighter graduate hiring coexist with no measurable aggregate lift.1 Markets priced in a productivity surge. The data, so far, does not support it.
Without that dividend, central banks navigating political and leadership uncertainty have no AI-based justification for patience on rates. Structural inflation — energy, services, supply chains — remains the dominant force. The productivity revolution is still a forecast.