Undercurrent CapitalbetaSign in

AI Labs Say Their Models Are Speeding Up AI Research

TLDR — METR estimates Anthropic's benchmark progress has nearly doubled in pace.

Signal data─ Neutralconfidence 50%Low materiality
AssetEventresearchEvent timeAttention0 readers/hrNormalSmart readers0

METR estimates Anthropic's benchmark progress has nearly doubled in pace. OpenAI reports higher estimated task-success rates without human intervention, with wide variation by task.... Read the full story at The Defiant

0 impressions · 0 upvotes · 0 comments

Reading is open to everyone — posting, replying and reporting need an account.

No comments yet.