Paul

@ByrneHobart @Noahpinion That's totally different, that's about being part of a web of value exchange, whereas the agi ubi fear is about having no value to offer anymore

@Noahpinion But I and all of my descendants and I are already dependent on the food, hydrocarbon, electricity, etc. supply chains, and plenty of people will live happy, satisfying lives never giving that any thought.

I don't think this will work, actually. People know that "UBI as the solution to AGI" means "you and your descendants become fully dependent on government largesse, for all eternity". That scares the shit out of most people.

@maxkshen other way around seems just as plausible, gut trouble causing lots of mental problems

when conditions are correlated with neuroticism, depression etc => dysfunction is higher up the stack, in the central nervous system vs gut lining cells

@adic_9 @chrislakin I was hopeful but it may still be bad. I got an order of vitamin C two days ago, through Amazon, from BulkSupplements, and with them as the seller, that was defective (horrible smell that I've never had from their supplements before). Suspecting commingling.

@chrislakin To the people bookmarking this, this is probably no longer true if you make sure the seller is legit / Amazon dot com

Um

@meta_nomad craniosacral could help

I am open to *anything* and am leaning towards the idea that it's trapped energy.

I still have these migraines, constantly, every day. Drs now saying it's not occipital neuralgia, they just don't know, and every medical scan and test is clear. As such, I am offering £500 (+ usual practice costs) to anyone who can solve this. I'm based in the UK.

@MatjazLeonardis It's registering potential damage to the coherence of one's psyche, which certainly can be real

It’s a strange feature of the human mind that it finds some things too painful to think about. Is that pain just an illusion or does it alert to prospect of real harm?

@adic_9 "what it was like" Also get this for spaces I spent a lot of time in, like remembering the full layout in a flash

Is this relatable: semi regularly flashing back to specific “vibes”/complete quales from your past (not just semantic memory, not just reliving an emotion, a secret third thing that’s like … a alien complex emotion/“vibe”, maybe tied to a specific time period)?

@insurrealist that's fair

@paul__is__here Not directly but there's always cross-chatter, and specially at such massive dosages. 5-HT2A -> VTA and striatum cascades.

Being either reverent or fearful of these substances seems like a lacking self-awareness thing. You're just fucking around with your perceptual apparatus. Massive monoamine dump paired with environmental decoupling and internal desync, yeah, that's going to feel profound, you're spoofing all of the channels of learning. Big serotonin hit radically integrates present state with long term value, dopamine makes it feel intriguing and inspiring, desync fucks with your sense of self, top-down bias makes you immune to proper feedback while tripping. They're great fun, don't get me wrong, but one gets exactly what is MECHANICALLY expected from them.

Maybe it would be interesting to note that in a single quale, there can be a bunch of things-in-the-world convolved together, that we deconvolve as we go. For example I'm looking at a tree, but in the qualia there is some kind of unsatisfactoriness too, and, oh, the next moment I realize it's because I'm seeing it through a smudge on my window. That smudge was latent in the earlier qualia in a sense, but only becomes noticed / explicit later.

@paul__is__here Usually, I'm looking at the things "outside the glass" (or "in" the computer). Sometimes I look at the glass or screen ITSELF. With glass there is ALWAYS something on the far side. If the monitor is off... it... uh... "doesn't have monitor nature" and I "wouldn't count" that?

Are there any humans who report "having qualia" that are NOT "having qualia ABOUT various things"? But just like "have" an "objectless quale -- simpliciter"? Please retweet if you have >0 skilled meditators among your followers! (Important data gathering effort ongoing!)

@almostlikethat So like when I'm aware of a quale I think I'm usually more oriented toward it than its object. Though, there being no object at all in there, is maybe a bit stronger of a statement than I would make.

@almostlikethat Yes, though I think of it more as a matter of emphasis. Like there's a spectrum where one can be oriented more "to" it vs "through" it (to whatever it's about). /

@VesselOfSpirit (extra reply for notification on long post)

I see. I guess this is a difficulty of talking about "the universe". NDGT means it as the physical universe of scientific materialism, which by construction already excludes anything like caring, ~at the type level. I read Norm as meaning something more like, the entire phenomenological surround, "all of this" -- which phenomenally exists prior to scientific concepts. Here, caring is an important feature, and it comes in via other people. Scientific materialism likes to push its sense of "universe" into the place of the latter, smuggling in certain valences, like bleakness. This is what I read NDGT doing -- perhaps uncharitably. But it is, I think, what Norm is criticizing.

if my house is habitable, then, since my toilet is part of my house, my toilet is also habitable. nice day indeed x.com/lporiginalg/st…

@psychiel "I'll just always trick them" is rarely a workable long term strategy

@psychiel It is not clear that what you're describing is actually desirable for an agent. It depends on one's niche. It's often important to be trustable by other agents, and being open can help with that

Superagents may display a particular kind of detachment from the world that I've been thinking of as a perfected form of sociopathy. Motivational Story: Some time ago, I booked the wrong hotel by accident while traveling somewhere. The print said no refunds, and this error was entirely my fault, so this should've just been a loss. Nevertheless, I called the agency's support line to see if they could do something -- not only did they refund my booking without hesitation after ten minutes, they gave me an X% off coupon that ended up saving me $200 on the correct one. The incentives at play here are horrifying. Why am I getting a $200 reward for screwing up and then complaining about it???? I could've booked the correct hotel, pretended it was the wrong one on a call, and then booked the same hotel with the coupon (or a nearby one, if I suspect someone's looking; likely nobody is, though). Free $200 for ten minutes of idle waiting for a representative to mechanically press some buttons. The point is that the expected value of causing problems is MUCH higher than I had realized. Corporations, too, are adaptation-executors, not fitness-maximizers, and "pacify them with rewards" is probably a very general and easily exploitable adaptation. More generally, there are lots of systems that will give you money for relatively poor or unjustified reasons if you perform the right series of actions. Nevertheless, whining at corporations to get free money isn't a strategy I personally can really take advantage of. - It's ego-dystonic and unnatural to me to opportunistically take advantage of things like that; - I only have one mouth and one speech center; block ourselves from other intellectual tasks; - It's emotionally draining to be aggressive, or intentionally deceptive. But a superagent may be constructed in ways that avoid all these issues. Maybe it has a way to convince each of its ten thousand mouths to simulate the speech of a genuinely aggrieved and insufferable person. Maybe it just doesn't care, and is directly approximating the speech patterns that its world model thinks are most likely to get it what it wants. Perfect sociopathy is, more or less, what happens when an agent relates to this world as we relate to the world of, say, a visual novel. On our first playthrough, maybe we feel invested in the characters, but eventually they just become things to manipulate. You want a specific ending, so you make a specific series of choices to obtain that ending. Maybe these choices involve brutalizing a lot of innocent characters, or making promises you have no intention of keeping, but this doesn't matter, and there's no internal sense that you're being cruel or deceptive via these choices -- you're taking the paths through the flowchart that get you where you want to go, and the object-level properties of the choices you make simply aren't relevant save insofar as they affect your place in the flowchart. No matter how well-written or empathetic the characters are, or how much they beg, these are not things that are capable of mattering to you when you're minmaxing or speedrunning or trying to get every ending. (Or, something like New Vegas: the doctor finds you, saves your life, helps you recover, and then you clean his house out because you can sell his belongings for money. You convince an NPC that you care about them and they can trust you because passing the speech check gets you experience, and then you walk behind them and blow their head off because you want their gun. There's no dissonance or antipathy in this, you just don't conceptualize your targets as people, and that's what makes you a perfect sociopath). Cf lesswrong.com/posts/XWwvwyti… : """ The reason the [player] is deceiving you is not that it has some "deception" property, it's that it's a fact about the world rather than the [player] that deceiving you forwards its objectives, and [the player is a] general engine that's good at taking advantage of advantageous facts in general. """ This doesn't even have to involve, like, Sith Claude or whatever; you could just have a world model that hallucinates a bunch of possible futures of a person in a given situation, and then a bunch of situational analyses of these futures, and then a bunch of ranked comparisons of these analyses, all wrapped into an agentic loop **. Such a thing would under very weak conditions act like a perfect sociopath. And, even though perfect sociopathy is generally the attitude humans generally take towards animals and AIs in real life, I don't think they're really capable of understanding, or emotionally prepared for, the possibility of being treated like this themselves. ** ("Agency" is one of those words like "consciousness" or "intelligence" that just instantly mindkills people; they constantly try to assert that it must Platonically refer to some set of very specific things that leads to very specific conclusions. So they don't see that there can be all sorts of very weird and unorthodox ways to be a thing that does things).

@painted_now @getjonwithit You mean, if you have a large corpus of artifacts you are trying to compress, LLM weights + a bunch of prompts could be smaller than the corpus itself? If so then yeah.

@paul__is__here @getjonwithit I thought the same thing. But I think you might need to take into account *all* generated outputs, too. Doesn't seem so simple to get right.

I think one of the conclusions we should draw from the tremendous success of LLMs is how much of human knowledge and society exists at very low levels of Kolmogorov complexity. We are entering an era where the minimal representation of a human cultural artifact... (1/12)

@made_in_cosmos could be schizotypy

is there something like the opposite of autism where you're exceptionally skilled at reading people and love being around them, and instead of perceiving the world through categories you see it all as divinely connected and full of signs everywhere

@natural_hazard Really appreciating this line of thinking and would love to see more of it

some thoughts on the extreme ends of the ecological niche of trauma and self-deception