Skip to content
Safety

Jev as a CoT Monitor: 6x Faster and 500x Cheaper.

LessWrongAnalysis or commentary··Updated yesterday
AI brief

A post describes Jev, a new model format that outputs certainties over a defined set of options instead of text, and applies it to monitoring harmful thought traces.

Why it matters: The post claims this classification-based monitoring approach is faster and cheaper, which could make monitoring harmful reasoning traces more practical.

Written by AI from LessWrong's published text. Read the original for full details.

Source

Interpretation or community commentary rather than straight reporting.

Read original story ↗