i might be confused about this but “witnessing a super-early universe” seems to support “a typical universe moment is not generating observer moments for your reference class”. but, yeah, anthropics is very confusing, so i’m not confident in this.
three most convincing arguments i know for OP’s thesis are:
atoms on earth are “close by” and thus much more valuable to fast running ASI than the atoms elsewhere.
(somewhat contrary to the previous argument), an ASI will be interested in quickly reaching the edge of the hubble volume, as that’s slipping behind the cosmic horizon — so it will starlift the sun for its initial energy budget.
robin hanson’s “grabby aliens” argument: witnessing a super-young universe (as we do) is strong evidence against it remaining compatible with biological life for long.
that said, i’m also very interested in the counter arguments (so thanks for linking to paul’s comments!) — especially if they’d suggest actions we could take in preparation.
i would love to see competing RSPs (or, better yet, RTDPs, as @Joe_Collman pointed out in a cousin comment).
Sure, but I guess I would say that we're back to nebulous territory then—how much longer than six months? When if ever does the pause end?
i agree that, if hashed out, the end criteria may very well resemble RSPs. still, i would strongly advocate for scaling moratorium until widely (internationally) acceptable RSPs are put in place.
I'd very surprised if there was substantial x-risk from the next model generation.
i share the intuition that the current and next LLM generations are unlikely an xrisk. however, i don't trust my (or anyone else's) intuitons strongly enough to say that there's a less than 1% xrisk per 10x scaling of compute. in expectation, that's killing 80M existing people -- people who are unaware that this is happening to them right now.
the FLI letter asked for “pause for at least 6 months the training of AI systems more powerful than GPT-4” and i’m very much willing to defend that!
my own worry with RSPs is that they bake in (and legitimise) the assumptions that a) near term (eval-less) scaling poses trivial xrisk, and b) there is a substantial period during which models trigger evals but are existentially safe. you must have thought about them, so i’m curious what you think.
that said, thank you for the post, it’s a very valuable discussion to have! upvoted.
the werewolf vs villager strategy heuristic is brilliant. thank you!
if i understand it correctly (i may not!), scott aaronson argues that hidden variable theories (such as bohmian / pilot wave) imply hypercomputation (which should count as an evidence against them): https://www.scottaaronson.com/papers/npcomplete.pdf
interesting, i have bewelltuned.com in my reading queue for a few years now -- i take your comment as an upvote!
myself i swear by FDT (somewhat abstract, sure, but seems to work well) and freestyle dancing (the opposite of abstract, but also seems to work well). also coding (eg, just spent several days using pandas to combine and clean up my philanthropy data) -- code grounds one in reality.
having seen the “kitchen side” of the letter effort, i endorse almost all zvi’s points here. one thing i’d add is that one of my hopes urging the letter along was to create common knowledge that a lot of people (we’re going to get to 100k signatures it looks like) are afraid of the thing that comes after GPT4. like i am.
thanks, everyone, who signed.
EDIT: basically this: https://twitter.com/andreas212nyc/status/1641795173972672512
AI takeover.