It’s been just over three years since ChatGPT launched—and maybe roughly three minutes since the last viral warning about an imminent white-collar “bloodbath.” Of course, as has been correctly said, it’s tough to make predictions, especially about the future.
So what does the data actually say so far? A growing stack of careful studies points in a remarkably consistent direction. Artificial intelligence is indeed affecting the US job market—as a powerful general-purpose technology should, at some point—but in ways that are narrower and slower than the apocalypse narrative suggests.
Start with timing. One recent paper finds that labor-market weakening in AI-exposed occupations began in early 2022—months before ChatGPT’s release. The authors argue this pattern points to broader macroeconomic and sectoral forces, including monetary tightening and a post-pandemic correction in tech hiring. That, rather than a clean break caused by generative AI alone. At the same time, AI-related skills such as writing, coding, and information synthesis remain valuable, with graduates exposed to them earning higher pay and finding jobs faster.
Another widely cited study, “Canaries in the Coal Mine” from the Stanford Digital Economy Lab, finds clearer AI-linked effects after careful statistical controls. The strongest signal emerges beginning around 2024 and is concentrated primarily among young workers in highly exposed occupations. Even there, the impact appears mainly to reflect reduced hiring rather than layoffs, with firms likely trimming junior inflows rather than displacing experienced staff.
Other analyses tell a similar story. According to the Dallas Fed, employment declines linked to AI exposure are small and concentrated among workers ages 20–24, driven mainly by reduced entry into those occupations rather than layoffs. The aggregate impact on overall unemployment remains minimal.
The big picture looks even less threatening. A recent analysis by The Economist notes that the US economy has added millions of white-collar jobs since late 2022. Real wages have risen. There has also been strong growth in technical, managerial, and coordination-heavy roles, including many often labeled AI “at risk.”
Put it all together and the state of play seems clear here in mid-February 2026. AI’s early job-market effects look like a shift at the entry level, at worst, not mass displacement. And one might reasonably conclude that hiring is slowing in routine knowledge roles even as demand rises for experienced, judgment-intensive, and AI-complementary work. At least the first phase of the Great AI Labor Transition appears evolutionary, not revolutionary—and certainly not catastrophic.
To quote the Stanford team, “We will be actively monitoring the labor market and updating our results periodically to see if the trends we identified persist, strengthen, reverse, or change in some other ways.”
Same with me.