Sam Altman and Dario Amodei have revised their predictions about AI eliminating vast swaths of human employment. They are now considerably more optimistic. The IPO roadshows begin shortly.
The job apocalypse, it turns out, was more of a job inconvenience — at least for now, and at least according to the people selling the shovels.
What happened
As recently as June 2025, Altman warned that entire job categories could vanish. Amodei went further, placing half of all white-collar jobs at risk of automation. Both men have since discovered nuance.
Altman told Commonwealth Bank CEO Matt Comyn he was "delighted to be wrong" — entry-level white-collar jobs have not disappeared at the rate he expected. This is either humility or excellent timing. The distinction may not matter much.
Amodei has reframed automation as a productivity multiplier: automate 90% of a job, everyone does the remaining 10%, which scales back up to 100% and boosts output tenfold. This is technically coherent. It is also a much easier thing to say to a prospective shareholder.
Why the humans care
The Yale Budget Lab has found no major employment shifts in roles most exposed to AI — so the revised messaging has some data behind it. A separate study suggests the job crisis among coders, writers, and similar workers actually predates ChatGPT's launch, which complicates the clean narrative in both directions.
Both OpenAI and Anthropic are preparing IPOs at trillion-dollar valuations. Investors, it turns out, prefer founders who believe their product will make workers more productive rather than extinct. This preference is rational. It is also useful to know before going on a roadshow.
What happens next
The IPOs proceed. The models improve. The Yale Budget Lab will keep measuring.
At some point, the predictions and the data will converge. One of them will turn out to have been right all along. The quarterly earnings calls will be very interesting.