My credence in N&E being basically right is more than Matt, but less than 50% What I really like about the book, though, is that it's great for skeptics. Most things are said clearly and with enough precision to make it easy to notice which parts are convincing and which are not

I thought about reposting my mini review from FB on Twitter, but it seemed like a hassle. Fortunately, Rob did it for me (and I of course appreciate the praise, Rob Bensinger ⏹️)

This is about 30% the same as the thread I started writing in response to Will's review just now. The main thing I would add is that I don't think the IABIED story was all *that* fast on take-off, and I don't think it would be too hard to rewrite for slower takeoff anyway.

Lots of weird reactions here like "Why are you complaining about normal political tactics?", "Why do you think this is unfair, when AI safety has X dollars?", and "What did you expect when you spent X on regulation?" Who said anything about this being unfair or unexpected??

#7 Combined Print & E-Book Nonfiction #8 Hardcover Nonfiction

This was a good post and you should read it. But a nitpick: The problem is the question, not the public, and calling it "trivial" vs "practical" is unfair. The danger from spent fuel is a solved problem, but it's totally reasonable to call it "dangerous" (tho perhaps not 'very')

AI is driving people to psychosis Our op-ed in today’s The Wall Street Journal explains why AI companies are incentivized to make the problem worse 🧵

I got a very convincing AI call trying to get me to schedule a follow-up appointment with my dentist a few minutes ago. What a bizarre world we're entering.

One time, I poured coffee from a carafe so badly someone said it looked one of those videos of poorly functioning robots making messes.

Technological innovation can be a form of participation in the divine act of creation. It carries an ethical and spiritual weight, for every design choice expresses a vision of humanity. The Church therefore calls all builders of #AI to cultivate moral discernment as a

Interesting overview on the loan guarantee thing from Dean Ball. I'd be curious if anyone who is similarly plugged-in thinks he's missing something. Also, I really appreciate that he answers naive questions from people like me in the replies when he posts things like this!

Seems like a majority of papers on governing AI are either: 1. Position papers explaining why using governance to constrain AI is a fool's errand or why the "real problem" is the author's hobby horse 2. Frameworks/models which are either too high-level to be useful or secretly 1

Presumably the "official" counterargument is that AI development is needed for natsec reasons and state regulations inhibit AI development nationally. I don't buy it, but I'd be interested if anyone has examples of state regulations that inhibited natsec tech in the past.

It's been obvious to me for a long time that one of the biggest unknowns about how AI will go is how much being very good at thinking lets you impact the world and achieve goals. People have *really* strong intuitions about this, which are, from what I've seen, never justified.