Some recent news, all long anticipated by this newsletter:
LLMs can still cannot be trusted:
A large fraction of what LLMs do is mostly just memorization (and Hinton was on the wrong side of this argument):
They still aren’t adding a lot of quantifiable value to the world:
Update: This is consistent with the finding of the Remote Labor Index that AI could only do about 2.5% of jobs, reported recently by the Washington Post.
Scaling isn’t going all that well, anymore, and probably won’t cure these problems.
Trying to orient our economy and geopolitical policy around such shoddy technology — particularly on the unproven hopes that it will dramatically improve– is a mistake.