LessWrong 2.0 Reader

Well, it was a bummer that my research on fasting found out I wasted my time doing 5:2 fasting for the last decade. Welp, I'll just research the next blog on calorie restriction. Everyone knows that's grea...

https://www.unaging.com/calorie-restriction/

Dammit. Why did I waste those five years doing calorie restriction before I started fasting?

Out of curiosity - “it's because Dustin is very active in the democratic party and doesn't want to be affiliated with anything that is right-coded” Are the projects related to AI safety or just generally? And what are some examples?

Steve writes:

the homunculus is definitionally the thing that carries “vitalistic force”, and that does the “wanting”, and that does any acts that we describe as “acts of free will”

Who wants things? Me of course. I want things. So the homunculus seems to be myself.

Google Gemini uses a watermarking system called Synth ID that claims to be able to watermark text by skewing its probability distribution. Do you think it’ll be effective? Do you think that it’s useful to have this?

Semi-related ACX post that came out today: Against The Cultural Christianity Argument.

(I don't think all silliness is fun. I have been hearing lame jokes of the same kind for months and they drive me up the wall.)

Basically I just agree with what James said. But I think the steelman is something like: you should expect shorter (or no) pauses with an RSP if all goes well, because the precautions are matched to the risks. Like, the labs aim to develop safety measures which keep pace with the dangers introduced by scaling, and if they succeed at that, then they never have to pause. But even if they fail, they're also expecting that building frontier models will help them solve alignment faster. I.e., either way the overall pause time would probably be shorter?

It does seem like in order to not have this complaint about the RSP, though, you need to expect that it's shorter by a lot (like by many months or years). My guess is that the labs do believe this, although not for amazing reasons. Like, the answer which feels most "real" to me is that this complaint doesn't apply to RSPs because the labs aren't actually planning to do a meaningful pause. 

Similar points apply [LW · GW] to adding too many [LW · GW] unnecessary links. Specifically links where it isn't clear where they lead and what point is made in the link target, as in the previous sentence.

This used to be a recurring failure mode of my own writing, which I've since partially mitigated. Reflecting on why, I think I wanted to do some combination of 

• justifying contentious or surprising claims 
• preventing being pattern-matched to straw versions of ideas / arguments commonly referenced and adjacent in concept-space 
• finding excuses to share cool reads
• making sense of links I'd read by relating them, using publication as a focusing mechanism 

I didn't notice the cost of overdoing it until I saw writers who did it worse, and became horrified at the thought that I was slowly becoming them.

(Gwern links a lot but it doesn't feel "worse", on the contrary I enjoy his writing, so "worseness" is as much about adding more value to the reader than the cost of disrupting their flow as it is about volume. His approach is also far more thought-out [LW(p) · GW(p)] of course.)

Good point!

Man, my model of what's going on is:

• The AI pause complaint is, basically, total self-serving BS that has not been called out enough
• The implicit plan for RSPs is for them to never trigger in a business-relevant way
• It is seen as a good thing (from the perspective of the labs) if they can lose less time to an RSP-triggered pause

...and these, taken together, should explain it.

Yeah, I think it's anti-natural for such an entity to exist, and that's a crux for religion-according-to-Alex.

Fantastic. I would love to figure out of this is true. 

It's also a crux for my worldview too. A Double Crux!