Post

Why does @techmeme.com continue to distribute these blatant and outrageous lies, told by Anthropic's P.R. department? I'm about to cross Techmeme off my list as "unreliable, full of lies"

Anthropic's index uses a fixed basket of tasks and model-assigned ratings, with no measured category fully autonomous. Independent replication next quarter would help distinguish changing task supervision from a proven increase in research output.

Source: Anthropic