Pee Doom

Karma: 717

Pee Doom 1 Feb 2024 19:38 UTC
12 points
2
on: Managing risks while trying to do good
How do we do this without falling into the Crab Bucket problem AKA Heckler’s Veto, which is definitely a thing that exists and is exacerbated by these concerns in EA-land? “Don’t do risky things” equivocates into “don’t do things”.

Pee Doom 14 Dec 2023 1:51 UTC
3 points
−2
in reply to: NicholasKross’s comment on: We’re all in this together
A medieval peasant would very much disagree with that sentence, if they were suddenly thrust into a modern grocery store. I think they would say the physical reality around them changed to a pretty magical-seeming degree.
They would still understand the concept of paying money for food. The grocery store is pretty amazing but it’s fundamentally the same transaction as the village market. I think the burden of proof is on people claiming that money will be ‘done away with’ because ‘post-scarcity’, when there will always be economic scarcity. It might take an hour of explanation and emotional adjustment for a time-displaced peasant to understand the gist of the store, but it’s part of a clear incremental evolution of stores over time.
- They think a friendly-AGI-run society would have some use for money, conflict, etc. I’d say the onus is on them to explain why we would need those things in such a society.
I think a basically friendly society is one that exists at all and is reasonably okay (at least somewhat clearly better) compared to the current one. I don’t see why economic transactions, conflicts of all sorts, etc wouldn’t still happen, assuming the lack of existentially-destructive ones that would preclude the existence of such a hypothetical society. I can see the nature of money changing, but not the fundamentals of there being trades.
I don’t think AI can just decide to do away with conflicts via unilateral fiat without an enormous amount of multipolar effort, in what I would consider a friendly society not ran by a world dictator. Like, I predict it would be quite likely terrible to have an ASI with such disproportionate power that it is able to do that, given it could/would be co-opted by power-seekers.
I also think that trying to change things too fast or ‘do away with problems’ is itself something trending along the spectrum of unfriendliness from the perspective of a lot of humans. I don’t think the Poof Into Utopia After FOOM model makes sense, that you have one shot to send a singleton rocket into gravity with the right values or forever hold your peace. This thing itself would be an unfriendly agent to have such totalizing power and make things go Poof without clear democratic deliberation and consent. This seems like one of the planks of SIAI ideology that seems clearly wrong to me, now, though not indubitably so. There seems to be a desire to make everything right and obtain unlimited power to do so, and this seems intolerant of a diversity of values.
This seems to be a combo of the absurdity heuristic and trying to “psychoanalyze your way to the truth”. Just because something sounds kind of like some elements of some religions, does not make it automatically false.
I am perfectly happy to point out the ways people around here obviously use Singularitarianism as a (semi-)religion, sometimes, as part of the functional purpose of the memetic package. Not allowing such social observations would be epistemically distortive. I am not saying it isn’t also other things, nor am I saying it’s bad to have religion, except that problems tend to arise. I think I am in this thread, on these questions, coming with more of a Hansonian/outside view perspective than the AI zookeeper/nanny/fully automated luxury gay space communism one.

Pee Doom 7 Dec 2023 23:15 UTC
5 points
−2
on: We’re all in this together
Nitpicking a particular topic of interest to me:
Power/money/being-the-head-of-OpenAI doesn’t do anything post-singularity.
It obviously does?
I am very confused why people make claims in this genre. “When the Singularity happens, this (money, conflict, the problems I’m experiencing) won’t be a problem anymore.”
This mostly strikes me as magical, far-mode thinking. It’s like people have an afterlife-shaped hole after losing religion. The specific, real reality in front of you won’t magically suddenly change after an Intelligence Explosion and assuming we’re alive in some coherent state. Money and power are very, very likely to still exist afterwards, just in a different state that makes sense as a transformation of the current world.

Pee Doom 5 Nov 2023 22:52 UTC
8 points
1
on: EA orgs’ legal structure inhibits risk taking and information sharing on the margin
I will keep harping on that more people should try starting (public benefit) corporations instead of nonprofits. At least, give it five minutes’ thought. Especially if handwaves impact markets something something. This should be in their Overton Window, but it might not be because they automatically assume “doing good ⇒ charity ⇒ nonprofit”. Corporations are the standard procedure for how effective helpful things are done in the world; they are RLHF’d by the need to acquire profit by providing real value to customers, reducing surfacce area for bullshitting. I am not an expert here by any means, but I’m noticing the fact that I can go on Clerky or Stripe Atlas and spend a couple hours spinning up an organization, versus, well, I haven’t actually gone through with trying to incorporate a nonprofit, but the process seems at least 10x more painful based on reading a book on it and with how many people seek fiscal sponsorship. I’m pretty surprised this schlep isn’t talked about more. Having to rely on fiscal sponsorship seems pretty obviously terrible to me, and I hadn’t even considered the information-distortive effects here. I would not be caught dead being financially enmeshed with the EVF umbrella of orgs after FTX. From my naive perspective, the castle could have easily been a separate business entity with EVF having at least majority control?
(I just realized I’m on LessWrong and not EA Forum, and could have leaned harder into capitalismpunk without losing as many social points.)

Pee Doom 8 Oct 2023 1:30 UTC
11 points
5
in reply to: Roko’s comment on: Sam Altman’s sister, Annie Altman, says Sam has (severely) abused her
The wifi hacking also immediately struck me as reminiscent of paranoid psychosis. Though a significant amount of psychosis-like things are apparently downstream of childhood trauma, including sexual abuse, but I forget the numbers on this.

Pee Doom 4 Oct 2023 6:24 UTC
4 points
0
in reply to: Tristan Williams’s comment on: EA Vegan Advocacy is not truthseeking, and it’s everyone’s problem
I’ve worried about it’s sustainability, but do you think it’s been a good path for you?
Cutting out bird and seafood products (ameliatarianism) is definitely more sustainable for me. I’m very confused why you would think it’s less sustainable than, uh, ‘cold turkey’ veganism. “Just avoid chicken/eggs” (since I don’t like seafood or the other types of bird meat products) is way easier than “avoid all meat, also milk, also cheese”.

Pee Doom 29 Sep 2023 21:47 UTC
2 points
1
in reply to: Kaj_Sotala’s comment on: Petrov Day Retrospective, 2023 (re: the most important virtue of Petrov Day & unilaterally promoting it)
Similar for me. I was very suspicious at first that the first message was a Scam and if I clicked I would blow up the website or something tricksy. Then with the second message I thought it might be customized to test my chosen virtue, “resisting social pressure”, so I didn’t click it.

Pee Doom 22 Aug 2023 8:43 UTC
−1 points
on: Dony’s Shortform Feed
“You’ve never experienced bliss, and so you’re frantically trying to patch everything up and pin it all together and screw the universe up so that it’s fixed.”—Alan Watts

Pee Doom 8 Aug 2023 4:53 UTC
1 point
in reply to: Annapurna’s comment on: Hedging the Possibility of Russia invading Ukraine
Did you actually bet the money?

Pee Doom 31 Jul 2023 20:43 UTC
2 points
1
on: Why was the AI Alignment community so unprepared for this moment?
People in MIRI/CFAR/LessWrong ~actively resisted the idea of a marketing push optimized more along dimensions of mass persuadability, for better or worse. One reason is that there is inertia once you’ve built a mass movement with MoPs who can’t dialogue like on this site. My straw model is they think “we just need to produce technical insights and communicate them” and other comms work is an opportunity cost or creates/incentivizes some kind of epistemic trap.

Pee Doom 31 Jul 2023 20:19 UTC
3 points
0
in reply to: Misaligned-Semi-intelligence’s comment on: Is Kennedy a Nazi?
And this is the exact type of statement where I would expect this coincidence to pop up. Both are a reasonable number of days for bureaucracy to take, the large discrepancy between them is required for the complaint in the tweet to happen in the first place, and I would expect the number of days to be very specific rather than a round number.
I agree with this. Although I will note people are claiming it actually took 56 (IIRC) days for them to get back to him.

Pee Doom 31 Jul 2023 20:16 UTC
1 point
0
in reply to: frontier64’s comment on: Is Kennedy a Nazi?
I think it’s an entirely sensible inference in the world where it is true that it was intentional, albeit a highly reductive description of the actual psychological reality of someone who holds those beliefs enough to output such symbols. In that world, he could also be paying lipservice to pander or be trolling.

Pee Doom 29 Jun 2023 19:33 UTC
3 points
2
in reply to: frontier64’s comment on: My tentative best guess on how EAs and Rationalists sometimes turn crazy
group homes (in adulthood as a long term situation)
People living together in group homes (as extended families) used to be the norm? The weird thing is how isolated and individualist we’ve become. I would argue that group houses where individual adults join up together are preserving some aspect of traditional social arrangement where people live closely, but maybe you would argue that this is not the same as an extended family or the lifelong kinship networks of a village.

Pee Doom 27 Jun 2023 22:29 UTC
1 point
on: Dony’s Shortform Feed
Has the “AI Safety”-washing grift accelerated? https://www.lakera.ai/momentum

Pee Doom 30 Jan 2023 6:33 UTC
2 points
0
on: “Status” can be corrosive; here’s how I handle it
Also, when I actually think about what it would be like to have a Plan That I Truly Believe In, it doesn’t really seem hard to get it on the desk of The Important Stakeholders (or someone who knows someone who talks to The Important Stakeholders).
I think you can scream a Good Plan from the rooftops and often few will listen. See: Covid in January-February 2020.

Pee Doom 6 Jan 2023 9:05 UTC
1 point
on: Dony’s Shortform Feed
Math camps are not sufficient to solve civilization’s problems.

Pee Doom 5 Jan 2023 3:03 UTC
6 points
3
on: Ritual as the only tool for overwriting values and goals
Fascinating. Way, way more examples and empirical treatment of rituals would help me understand your case better.

Pee Doom 31 Dec 2022 6:32 UTC
6 points
0
on: What do you imagine, when you imagine “taking over the world”?
Me, sitting on a throne, as your entirely benevolent world dictator. Oh, how did I get there? Someone posted on LessWrong and I followed their blueprint!

Pee Doom 26 Dec 2022 9:48 UTC
3 points
5
on: Necrobot
The story had me for most of it but I’m pretty disappointed by the very end. Protagonist seems to have unequivocally become the bad guy at that point. I don’t think I would be confident enough in my understanding of consciousness to do that until I spent a lot more time understanding the situation, assuming it wasn’t a self-defense emergency by that point. I get being angry but remaining in reflexively unthinking action hero mode rather than administering a battery of Turing Tests just seems wildly uncalled for. Maybe it’s supposed to read as morally ambiguous, but like many readers here I am biased towards the Drangerian perspective and find the dilemma unsympathetic.

Pee Doom 10 Nov 2022 5:05 UTC
3 points
−2
on: Trying to Make a Treacherous Mesa-Optimizer
One possible approach to fixing this is to try to get wayyyy more empirical, and try to produce proof-of-concept implementations of various adversaries we are worried we might face in the future. My analogy would be, there’s a world of difference between speculating about the bogey monster and producing a grainy photo of the bogey monster; the second can at least maaaaaybe be discussed with skeptical people, whereas the first cannot (productively) be.
Anyway, that’s a long-winded way of saying, it seemed to me that it might be useful to implement a treacherous mesa-optimizer in a toy grid-world, so I did. Here is the colab. Below I give a brief recap of the high-level results.
I’m not saying this isn’t helpful work worth the tradeoff, but this sounds like very early stage gain of function research? Have you thought about the risks of this line of investigation turning into that? :)