For Claude, capability and CDT are the ~same thing. Less so for GPT.

LessWrong reports on research examining two qualities in Claude and GPT models: raw capability and how well they handle a set of decision-theoretic problems. Researchers tested whether smarter models also solve these problems better. For Claude, the correlation was striking—point ninety-seven, nearly perfect. For OpenAI, it measured point fifty-five. The finding suggests Claude's and OpenAI's models scale differently: as Claude gets smarter, it also gets better at decision problems, while OpenAI's capabilities and problem-solving appear to improve more independently.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton