Arguments for P
ai
LessWrong has published an essay in which prominent AI researchers—including figures from OpenAI, DeepMind, Anthropic, and independent researchers—discuss their credence in something they refer to only as 'P,' without ever defining what it is. Contributors including Daniel Kokotajlo, Ryan Greenblatt, Joe Carlsmith, and others assign probabilities ranging from sixty to eighty percent, with Kokotajlo at eighty percent and Paul Christiano at sixty. The piece appears to be a meta-commentary on how sensitive or speculative ideas circulate within the AI research community through coded language and implication rather than explicit statement.
Source: https://www.lesswrong.com/posts/NG2AigxmBKLu9oCZE/arguments-for-p
Listen to this story
Hear this and more stories in a personalized audio briefing.
Open The Chonkerton