- A February 2026 NBER study of roughly 6,000 executives found nearly 90% of firms reported no AI impact on employment or productivity over the past three years.
- Among executives who reported using AI, average usage was just 1.5 hours per week; 25% reported no workplace AI use at all.
- Apollo chief economist Torsten Slok compared the pattern to Solow’s 1987 productivity paradox, noting AI is absent from employment, productivity, and inflation data.
- Competing academic studies show contradictory results: the St. Louis Fed measured a 1.9% cumulative productivity gain since ChatGPT‘s 2022 launch; a 2024 MIT study projected only 0.5% over the next decade.
What Happened
A study published in February 2026 by the National Bureau of Economic Research surveyed approximately 6,000 CEOs, chief financial officers, and other executives from firms participating in business outlook surveys across the United States, United Kingdom, Germany, and Australia. Nearly 90% of those firms reported that AI has had no impact on employment or productivity over the previous three years. The findings were reported by Fortune on February 17, 2026.
While roughly two-thirds of surveyed executives said they use AI at work, that usage averaged just 1.5 hours per week. One in four respondents reported no AI use in the workplace at all.
Why It Matters
Torsten Slok, chief economist at Apollo Global Management, drew a direct parallel to Solow’s productivity paradox — Nobel laureate Robert Solow’s 1987 observation that computing technology was failing to appear in productivity statistics despite widespread corporate adoption. “AI is everywhere except in the incoming macroeconomic data,” Slok wrote in a blog post. “Today, you don’t see AI in the employment data, productivity data, or inflation data.”
The NBER findings arrive as corporate AI investment exceeded $250 billion in 2024. A Financial Times analysis covering S&P 500 earnings calls from September 2024 through 2025 found 374 companies mentioned AI, with most characterizing their adoption as positive — yet those assessments have not translated into measurable aggregate output gains. Slok added that outside the Magnificent Seven technology companies, there are “no signs of AI in profit margins or earnings expectations.”
Technical Details
Existing academic research offers conflicting readings on AI’s productivity contribution. The Federal Reserve Bank of St. Louis, in its State of Generative AI Adoption report published in November 2024, documented a 1.9% increase in excess cumulative productivity growth since ChatGPT‘s introduction in late 2022. A separate 2024 MIT study reached a far more conservative estimate: 0.5% productivity growth projected over the next decade.
“I don’t think we should belittle 0.5% in 10 years. That’s better than zero,” said Daron Acemoglu, a Nobel laureate and MIT economist who authored the latter study. “But it’s just disappointing relative to the promises that people in the industry and in tech journalism are making.”
ManpowerGroup’s 2026 Global Talent Barometer, which surveyed nearly 14,000 workers across 19 countries, found that while regular AI use rose 13% in 2025, worker confidence in the technology’s utility fell 18%. A Boston Consulting Group study of 1,488 U.S. full-time workers found self-reported productivity increased when workers used one to three AI tools but declined when they used four or more, with respondents citing cognitive fatigue and increased errors.
Who’s Affected
The divergence between corporate AI spending and measured returns has direct implications for organizations that have committed capital to AI infrastructure. IBM chief human resources officer Nickle LaMoreaux announced in 2026 that the company would triple its intake of entry-level hires, citing concern that automating junior roles would deplete the pipeline feeding future management positions.
A Stanford Institute for Economic Policy Research study using internet browsing data from 200,000 U.S. households found generative AI improved efficiency on online tasks — including job hunting, travel planning, and shopping — by between 76% and 176%. Researchers found, however, that time freed up by AI use was spent on leisure rather than on work or skills development, limiting its measurable economic contribution.
What’s Next
Erik Brynjolfsson, director of Stanford University’s Digital Economy Lab, argued in a Financial Times op-ed that a reversal may already be underway, citing fourth-quarter GDP tracking at approximately 3.7% and estimating a U.S. productivity increase of 2.7% in 2025, which he attributed to companies beginning to extract returns on earlier AI investment. Former Pimco CEO Mohamed El-Erian similarly noted the decoupling of job growth and GDP growth as a structural signal comparable to the office automation shift of the 1990s.
Slok suggested AI’s economic trajectory could follow a “J-curve” — initial underperformance followed by an accelerating surge — but framed the outcome as contingent on deployment. “From a macro perspective, the value creation is not the product,” Slok wrote, “but how generative AI is used and implemented in different sectors in the economy.”
