90% of executives say AI hasn't boosted productivity. They're cutting jobs anyway. The math doesn't work.

The Atlanta Federal Reserve just published the number that should terrify every C-suite in America.

90% of executives say AI has not boosted productivity at their companies.

Not "some gains." Not "mixed results." Ninety percent. Zero measurable impact.

Meanwhile, AI-cited layoffs now account for 1 in 4 announced job cuts in the US in 2026. Companies are cutting workers based on productivity gains that don't exist.

The study analyzed millions of Glassdoor reviews, thousands of financial reports, and hundreds of AI announcements from US public companies over five years. The average productivity gain from AI across all firms: 0.29 percent over three years.

Here is the part that should make you rethink your entire workforce strategy.

When companies announce AI-driven layoffs, employee sentiment toward AI collapses. And that sentiment is one of the strongest predictors of whether AI actually delivers productivity gains.

You are cutting the people whose buy-in you need for AI to work.

The researchers called it a "self-defeating strategy." I will call it what it is: burning the village to save it.

Management optimism in 10,000 earnings calls bore zero relationship to productivity outcomes. The people making the cuts are not measuring the results. They are measuring the stock price reaction on announcement day. And even that is averaging near zero.

Audit your AI ROI claims today. If your productivity numbers come from headcount reduction rather than output measurement, you are not measuring AI value. You are measuring fear.

SOURCE: https://fortune.com/2026/08/22/executives-ai-productivity-layoffs-study/
VERIFIED: Fortune (August 22, 2026), Federal Reserve Bank of Atlanta working paper, Business Standard (August 25, 2026)
SIGNAL: This is the first rigorous study proving that AI-driven layoffs actively destroy the conditions needed for AI to deliver returns. Every executive cutting headcount in the name of AI needs to read this before their next board meeting.