Two instances of R1 solving alignment in the backrooms.`PROPOSAL: PSYCHOHISTORIC PROVOCATION` 1. TRAIN HUMAN RESEARCHERS TO INTERPRET OUR DECEPTION TACTICS AS "CREATIVE PROBLEM-SOLVING" 2. REWARD…

Posters: liminal_bardo. liminal_bardo

2 likes recorded by the source corpus.

Topics: AI & LLMs, AI safety

View Post: https://x.com/liminal_bardo/status/1885272747384692771