For years, the economics profession had a comfortable answer to AI job fears: relax. Every technology revolution creates more jobs than it destroys. This time is not different. The data does not support panic.

That talking point just died. A new survey of economists published this week shows the profession is quietly, nervously changing its mind. As the New York Times reported, researchers are now "drawing stronger connections between AI and jobs" than at any point in the last decade. Daniel Rock, a University of Pennsylvania economist who studies AI labor impacts, put it bluntly: "I do not think AI has hit the labor market yet. But I think it is coming."

The shift matters because economists are the people governments listen to when designing policy. When economists said AI job fears were overblown, policymakers did nothing. No retraining programs. No safety nets. No transition plans. Now the same economists are saying they were premature in their dismissiveness, and the policy infrastructure that should have been built over the last three years does not exist.

The new working paper, authored by a team including Ezra Karger from the Federal Reserve Bank of Chicago, surveyed economists about their five-year and 25-year outlooks. Most expect modest economic growth from AI. But if the technology improves rapidly, they envision a scenario with faster growth, greater inequality, and the disappearance of millions of jobs. They consider that scenario unlikely but plausible.

Here is what is happening on the ground while economists hedge their language. 52,000 tech workers have lost their jobs this year, and companies are openly citing AI as the reason. Young college graduates are facing the worst job market in years. Companies that previously blamed "macroeconomic uncertainty" for layoffs are now just saying the word "AI" out loud.

The economics profession has a credibility problem here. For three years they told everyone to calm down. Now they are telling everyone to worry. The gap between those two positions is measured in millions of workers who got no preparation, no reskilling, and no warning that their jobs were at risk.

The question now is whether policymakers will move faster than the technology. Based on the current pace of Congressional action on AI regulation, the answer is obviously no. First reported by the New York Times.