1+ hour, 46+ min ago  (229+ words) There are laws against recklessly endangering people's lives, and laws against creating a public nuisance. But, at least in the USA, the bar for bringing a case on such grounds seems to be quite high. Several people I've talked to…...

1+ hour, 43+ min ago  (1262+ words) (This was originally going to be a "quick take" but then it got a bit long. Just FYI.) There's this weird trend I perceive with the personas of LLM assistants over time. It feels like they're getting less "coherent" in…...

2+ hour, 36+ min ago  (453+ words) (Fragments from a research paper that will never be written) The frontier AI developers are becoming increasingly powerful and wealthy, significantly increasing their potential for risks. One concern is that of executive misalignment: when the CEO has different incentives and…...

5+ hour, 36+ min ago  (12+ words) Assuming transformers can have conscious experience, what would that experience be like? "...

7+ hour, 15+ min ago  (1059+ words) I've been playing with causal inference lately, as one does. [1] I was thinking of writing a more formal sequence about how to do causal discovery and...

8+ hour, 19+ min ago  (716+ words) Arguably, AI welfare work'is relatively non-urgent and can be left'until after the intelligence explosion, since it is hard to make progress on and not needed to avoid AI takeover or authoritarian lock-in. Here is a basic case for why we…...

8+ hour, 56+ min ago  (857+ words) A few days ago I saw this comic reposted, and I thought: wait! Unlike every prior time I have seen this comic, I actually know how to solve this now!...

9+ hour, 20+ min ago  (549+ words) Originally posted as a'comment on'this post. Reposting for visibility and since it is lengthy enough to be a standalone post. I plan to post a more comprehensive update in future describing FRI's impact and theory of change in more detail....

9+ hour, 38+ min ago  (306+ words) TL; DR. AI safety needs more people who understand the field and its gaps deeply enough to own problems end-to-end, found new projects and organizations, and shape the threat models that the rest of the field runs on. Astra is…...

10+ hour, 46+ min ago  (25+ words) Apart and the secure program synthesis community are launching a fellowship! A ton of you reading this would make great mentors" but the mentor appli...