Claude 4
product
7 mentions
6 recordings
first heard Sep 2024
last heard 29 Oct
↓4 vs the 12 months before
Claude 4, a large language model from Anthropic
Trend
Peak in the week of 23 Oct — 1 mentions across 1 series.
Moments
newest first · ▶ plays the momentNathan Labenz"The Cognitive Revolution" · Is AI Stalling Out? Cutting Through Capabilities Confusion, ... · 1:06:12 · 29 Oct
…And we have seen in the Claude Four and in the GPT five system cards, Claude Four reported, I think, a two-thirds reduction in…
Nathan Labenz"The Cognitive Revolution" · Full-Stack AI Safety: Why Defense-in-Depth Might Work, with ... · 38:34 · Sep 2025
…Um in the GPT-5 and in the Claude 4 report, there were also notable graphs showing like a two-thirds reduction in reward hacking behavior in in Claude 4 relative to 3.7, and GPT-5 had like a significant reduction in scheming relative to O three.…
Nathan Labenz"The Cognitive Revolution" · Zvi Mowshowitz on Longer Timelines, RL-induced Doom, and Why... · 1:31:47 · Sep 2025
…with Claude 4 on a reward ha internal reward hacking benchmark, they reported basically a two-thirds reduction.…
7 moments · sign in to read them all