Thinking Feed
Feed
Areas
AdminAgentic SoftwareAI Safety
You
AI-assisted AI research versus a self-sustaining recursive loop

Posts here

AI Safety · 2 min · 4 Oct

A frontier lab now has AI doing most of a substantial share of its research tasks, but none of the measured work runs without human oversight. Delegation is rising fast; whether it makes the next generation of AI improve itself faster remains unmeasured.

anthropic.com14 min
Measurements for understanding the pace of AI development inside frontier labs
30
AI Safety · 30 Sep

A short calibration for Amodei’s self-improvement worry: in September, METR found AI helping AI R&D but judged its tested model unlikely to automate research end to end. Real acceleration, not a demonstrated runaway loop; the missing test is whether gains compound across generations after accounting for humans and compute.

metr.org
metr.org
21