Wednesday, September 23, 2026

LLM slowdown

The "AI slowdown" is not caused by a coordinated reaction to safety concerns*. That's being used as an excuse. Rather, it is due to the terribly weak scaling law for LLMs. This was foreseen years ago; Ilya Sutskever's "end of the age of scaling" and return to another "age of research."**

* really they're doom trolling

** See also my blog of 1 January 2024.

No comments:

Post a Comment