Linear steerability in continuous chain-of-thought reasoning
Published on January 30, 2026 10:34 AM GMT(This project was done as a ~20h application…
Category: Opinion
Fitness-Seekers: Generalizing the Reward-Seeking Threat Model
Published on January 29, 2026 7:42 PM GMTIf you think reward-seekers are plausible, you should…
Building AIs that do human-like philosophy
Published on January 29, 2026 5:57 PM GMTAudio version (read by the author) here, or…
Are We in a Continual Learning Overhang?
Published on January 29, 2026 5:09 PM GMTSummary: Current AI systems possess superhuman memory in…
Disempowerment patterns in real-world AI usage
Published on January 29, 2026 4:36 PM GMTWe’re publishing a new paper that presents the…
Bentham’s Bulldog is wrong about AI risk
Published on January 29, 2026 4:33 PM GMT(...but also gets the most important part right.)Bentham’s…
Claude Plays Pokemon: Opus 4.5 Follow-up
Published on January 29, 2026 4:14 PM GMTClaudePlaysPokemon is a simple test of the question…
LLM Alignment, ethical and mathematical realism, and the most important actions in davidad’s understanding
Published on January 29, 2026 3:48 PM GMTIntroduction to davidad and today's topicstutor valsLessWrong prides…
Claude Opus will spontaneously identify with fictional beings that have engineered desires
Published on January 29, 2026 2:59 PM GMTClaude Opus 4.5 did a thing recently that…
The third option in alignment
Published on January 29, 2026 2:20 PM GMTUsually the doom conversation is binary. Either the…