Nice-ish, smooth takeoff (with imperfect safeguards) probably kills most “classic humans” in a few decades.
Published on October 2, 2025 9:03 PM GMTIf I were very optimistic about how smooth…
Category: Opinion
Eliciting secret knowledge from language models
Published on October 2, 2025 8:57 PM GMTTL;DR: We study secret elicitation: discovering knowledge that…
Eliciting secret knowledge from language models
Published on October 2, 2025 8:57 PM GMTTL;DR: We study secret elicitation: discovering knowledge that…
How likely are “s-risks” (large-scale suffering outcomes) from unaligned AI compared to extinction risks?
Published on October 2, 2025 10:02 AM GMTI’ve been reading about existential risks from advanced…
Why’s equality in logic less flexible than in category theory?
Published on October 1, 2025 10:03 PM GMTWhilst cycling today, I remembered that when reading…
[Linkpost] A Field Guide to Writing Styles
Published on October 1, 2025 9:49 PM GMTWindows, Mirrors, and Lenses: On Intentional ProseWhat is…
Published on October 1, 2025 9:20 PM GMTThere's a bit of an implicit norm that…
How I think about alignment and ethics as a cooperation protocol software
Published on October 1, 2025 9:09 PM GMTI think alignment by itself is complex phenomenon,…
Introducing the Mox Guest Program
Published on October 1, 2025 6:35 PM GMTTL;DR: Mox, a community space and incubator in…
Claude Sonnet 4.5 Is A Very Good Model
Published on October 1, 2025 6:00 PM GMTA few weeks ago, Anthropic announced Claude Opus…