x.com/gwern/status/2097892768219894221
1 correction found
their leadership believes they are about 6 months away from being, definitionally, able to fire all their human AI researchers and still develop AI.
Official statements from OpenAI and Anthropic contradict this. OpenAI said in June 2026 that its internal view was AI would do only a significant fraction of research by March 2028, in tandem with human researchers, while Anthropic said in August 2026 its models were not close to substituting for its research scientists and engineers.
Full reasoning
This claim is contradicted by recent official statements from the companies named in the post.
-
OpenAI: In an official June 8, 2026 post co-authored by CEO Sam Altman and Chief Scientist Jakub Pachocki, OpenAI wrote: "Our internal belief is that by March of 2028 we may have a significant fraction of our research being done by AI systems in tandem with our own researchers." That is not "about 6 months away" from September 2026, and it explicitly describes AI doing a significant fraction of research in tandem with human researchers—not leadership believing they will soon be able to fire all human AI researchers.
-
Anthropic: In Anthropic's official Transparency Hub update dated August 17, 2026, the company said none of 16 surveyed Anthropic researchers believed Claude could replace an entry-level researcher within three months, and elsewhere in the same report said the model was "not close to substituting for our research scientists and engineers." Anthropic also states its model did not fully automate even an entry-level, remote-only research role.
These are the strongest available direct statements of the companies' own views, and they point the opposite way from the post: both companies publicly describe AI as assisting or partially automating research, not making all human AI researchers fireable within roughly six months.
2 sources
- Built to benefit everyone: our plan | OpenAI
"Our internal belief is that by March of 2028 we may have a significant fraction of our research being done by AI systems in tandem with our own researchers."
- Anthropic's Transparency Hub | Anthropic
Anthropic says none of 16 surveyed researchers believed the model could replace an entry-level researcher within three months, and that the model is "not close to substituting for our research scientists and engineers."