I mean, Flynn Rider was also really good-looking. For a lot of people, maybe most, this look is just unattainable. Even if you can get in as good physical shape (which is far from easy), what if you're older, shorter, balder, have a goofy face and so on.

Yeah. I had a similar idea, that autism spectrum stuff comes from a person's internal "volume knobs" being turned to the wrong positions. Some things are too quiet to notice, while others are so loud that it turns into a kind of wailing feedback, like from a too loud microphone. And maybe some of it is fixable with exposure training, but not everything and not easily.

Wei's motivating example for UDT1.1 is exactly that. It is indeed weird that Eliezer's FDT paper doesn't use the idea of optimizing over input-output maps, despite coming out later. But anyway, "folklore" (which is slowly being forgotten it seems) does know the proper way to handle this.

I'm not sure question 1 is analogous to Newcomb's problem. We're trying to maximize the money made by the program overall, not just the money made in the specific (x,x) case. In other words, even if you're "inside" the problem, the UDT-ish thing to do is first figure out the optimal way to respond to all (x,y) pairs, and only then feed it the (x,x) pair. Wei Dai called this "UDT1.1", first optimizing over all input-output maps and then applying it to a specific input.

ideally the only 1/4″ cables onstage are short runs to DIs

And all the pedalboard stuff that happens before the DI. But mostly I agree.

Btw, do you already know that a piezo signal is much improved by a preamp with >1 meg ohm input impedance? I figured that out with my electric cello.

I think there's a worldwide trend toward more authoritarian leaders, which contributed to both these events. And it should raise our probability of e.g. Turkey or China doing something silly. But where this trend comes from, I'm not sure. It certainly predates the current AI wave. It could be due to social media making people more polarized or something. But then again there were plenty of worse dictators in history, long before social media or electricity. So maybe what's happening now is regression to the mean, and nice democracy was an anomaly in place and time.

Yeah. I remember where I was and how I felt when covid hit in 2020, and when Russia attacked Ukraine in 2022. This tariff announcement was another event in the same row.

And it all seems so stupidly self-inflicted. Russia's economy was booming until Feb 2022, and US economy was doing fine until Feb 2025. Putin-2022 and Trump-2025 would've done better for their countries by simply doing nothing. Maybe this shows the true value of democratic checks and balances: most of the time they add overhead, but sometimes they'll prevent some exceptionally big and stupid decision, and that pays for all the overhead and then some.

Your examples sound familiar to me too, but after rereading your comment and mine, maybe it all can be generalized in a different way. Namely, that internal motivation leads to a low level of effort: reading some textbooks now and then, solving some exercises, producing some small things. It still feels a bit like staying in place. Whereas it takes external motivation to actually move forward with math, or art, or whatever - to spend lots of effort and try to raise my level every day. That's how it feels for me. Maybe some people can do it without external motivation, or maybe they lucked into getting external motivation in the right way, I don't know.

I agree feedback is a big part of it. For example, the times in my life when I've been most motivated to play musical instruments were when I had regular opportunities to play in front of people. Whenever that disappeared, the interest went away too.

But also I think some of it is sticky, or due to personality factors. We could even say it's not about willpower at all, but about value differences. Some people are just more okay with homeostasis, staying at a certain level (which can be lower or higher for different people) and using only as much effort as needed for that. While others keep climbing and applying effort without ever reaching a level that lets them relax. Many billionaires seem to be of that second type. I'm more of the first type, with many of my active periods being prompted by external changes, threats to homeostasis. It's clear that type 2 achieves more than type 1, but it's not clear which type is happier and whether one should want to switch types.

Good post. But I thought about this a fair bit and I think I disagree with the main point.

Let's say we talk about two AIs merging. Then the tuple of their expected utilities from the merge had better be on the Pareto frontier, no? Otherwise they'd just do a better merge that gets them onto the frontier. Which specific point on the frontier is a matter of bargaining, but the fact that they want to hit the frontier isn't, it's a win-win. And the merges that get them to the frontier are exactly those that output a EUM agent, maximizing some linear combination of their utilities. If the point they want to hit is in a flat region of the frontier, the merge will involve coinflips to choose which EUM agent to become; and if it's curvy at that point, the merge will be deterministic. For realistic agents who have more complex preferences than just linearly caring about one cake, I expect the frontier will be curvy, so deterministic merge into a EUM agent will be the best choice.

"Apparatchik" in the USSR was some middle-aged Ivan Ivanovich who'd yell at you in his stuffy office for stepping out of line. His power came from the party apparatus. While the power of Western activists is the opposite: it comes from civil society, people freely associating with each other.

This rhetorical move, calling a Western thing by an obscure and poorly fitting Soviet name, is a favorite of Yarvin: "Let's talk about Google, my friends, but let's call it Gosplan for a moment. Humor me." In general I'd advise people to stay away from his nonsense, it's done enough harm already.

The objection I'm most interested in right now is the one about induced demand (that's not the right term but let's roll with it). Like, let's say we build many cheap apartments in Manhattan. Then the first bidders for them will be rich people - from all over the world! - who would love to get a Manhattan apartment for a bargain price. The priced-out locals will stay just as priced out, shuffled to the back of the line, because there's quite many rich people in the world who are willing to outbid them. Maybe if we build very many apartments, and not just in Manhattan but everywhere, the effect will eventually run out; but it'll take very many indeed.

The obvious fix is to put a thumb on the scale somehow, for example sell these cheap apartments only as primary residences. But then we lose the theoretical beauty of "just build more", and we really should figure out what mix of "just build more" and "put a thumb on the scale" is the most cost-efficient for achieving what we want. Maybe some thumb on the scale will even give us what we want without building more, since there's a lot of empty housing and non-primary housing.

Maybe you're pushing your proposal a bit much, but anyway as creative writing it's interesting to think about such scenarios. I had a sketch for a weird utopia story where just before the singularity, time stretches out for humans because they're being run at increasing clock speed, and the Earth's surface also becomes much larger and growing. So humanity becomes this huge, fast-running civilization living inside an AI (I called it "Quetzalcoatl", not sure why) and advising it how it should act in the external world.

My wife used to have a talking doll that said one phrase in a really annoying voice. Well, at some point the doll short-circuited or something, and started turning on at random times. In the middle of the night for example it would yell out its phrase and wake everyone up. So eventually my wife took the doll to the garbage dump. And on the way back she couldn't stop thinking about the doll sitting there in the garbage, occasionally yelling out its phrase: "Let's go home! I'm already hungry!" This isn't creative writing btw, this actually happened.

The thread about Tolkien reminded me of Andrew Hussie's writing process. Start by writing cool scenes, including any elements you like. A talking tree? Okay. Then worry about connecting it with the story. The talking tree comes from an ancient forest and so on. And if you're good, the finished story will feel like it always needed a talking tree.

I'd be really interested in a similar breakdown of JK Rowling's writing process, because she's another author with a limitless "toybox".

I think something like the Culture, with aligned superintelligent "ships" keeping humans as basically pets, wouldn't be too bad. The ships would try to have thriving human societies, but that doesn't mean granting all wishes - you don't grant all wishes of your cat after all. Also it would be nice if there was an option to increase intelligence, conditioned on increasing alignment at the same time, so you'd be able to move up the spectrum from human to ship.

Maybe tangential, but this reminded me of a fun fact about Hong Kong's metro: it's funded by land value. They put a station and get some land development rights near it. Well, building the station obviously makes land around it more valuable. So they end up putting stations where they'd be most useful, and fares can be cheap because the metro company makes plenty of money from land. So the end result is cheap, well-planned public transport which is profitable and doesn't take government money.

Not to pick on you specifically, but just as a general comment, I'm getting a bit worried about the rationalist book review pipeline. It seems it usually goes like this: someone writes a book with an interesting idea -> a rationalist (like Scott) writes a review of it, maybe not knowing much about the topic but being intrigued by the idea -> lots of other rationalists get the idea cached in their minds. So maybe it'd be better if book reviews were written by people who know a lot about the topic, and can evaluate the book in context.

Like, a while ago someone on LW asked people to recommend textbooks on various topics, but you couldn't recommend a textbook if it was the only one you'd read on the topic, you had to read at least two and then recommend one. That seems on the right track to me, and requiring more knowledge of the topic would be better still.

I think I can destroy this philosophy in two kicks.

Kick 1: pleasure is not one-dimensional. There are different parts of your brain that experience different pleasures, with no built-in way to compare between them.

When you retreat from kick 1 by saying "my decision-making provides a way to compare, the better pleasure is the one I'll choose when asked", here comes kick 2: your decision-making won't work for that. There are compulsive behaviors that people want to do but don't get much pleasure from them. And in every decision there's a possible component of that, however small.

You could say "I'll compare decisions based on how much pleasure they bring, excluding compulsiveness", but you can't do that due to kick 1 again. So the philosophy just collapses.

Good point. But I think the real game changer will be self-modification tech, not longevity tech. In that case we won't have a "slow adaptation" problem, but we'll have a "fast adaptation in weird directions" problem which is probably worse.

In Copenhagen every street has wide sidewalks and bike lanes in both directions, and there's lots of public transport too. It's good.

I don't understand Eliezer's explanation. Imagine Alice is hard-working and Bob is lazy. Then Alice can make goods and sell them to Bob. Half the money she'll spend on having fun, the other half she'll save. In this situation she's rich and has a trade surplus, but the other parts of the explanation - different productivity between different parts of Alice (?) and inability to judge her own work fairly (?) - don't seem to be present.

No. Committing a crime inflicts damage. But interacting with a person who committed a crime in the past doesn't inflict any damage on you.

Because the smaller measure should (on my hypothesis) be enough to prevent crime, and inflicting more damage than necessary for that is evil.

Because otherwise everyone will gleefully discriminate against them in every way they possibly can.

I think the US has too much punishment as it is, with very high incarceration rate and prison conditions sometimes approaching torture (prison rape, supermax isolation).

I'd rather give serial criminals some kind of surveillance collars that would detect reoffending and notify the police. I think a lot of such people can be "cured" by high certainty of being caught, not by severity of punishment. There'd need to be laws to prevent discrimination against people with collars, though.

Yeah, I stumbled on this idea a long time ago as well. I never drink sugary drinks, my laptop is permanently in grayscale mode and so on. And it doesn't feel like missing out on fun; on the contrary, it allows me to not miss out. When I "mute" some big, addictive, one-dimensional thing, I start noticing all the smaller things that were being drowned out by it. Like, as you say, noticing the deliciousness of baked potatoes when you're not eating sugar every day, or noticing all the colors in my home and neighborhood when my screen is on grayscale.

I suppose the superassistants could form coalitions and end up as a kind of "society" without too much aggression. But this all seems moot, because superassistants will anyway get outcompeted by AIs that focus on growth. That's the real danger.

I don't quite understand the plan. What if I get access to cheap friendly AI, but there's also another much more powerful AI that wants my resources and doesn't care much about me? What would stop the much more powerful AI from outplaying me for these resources, maybe by entirely legal means? Or is the idea that somehow the AIs in public access are always the strongest possible? That isn't true even now.

I also agree with all of this.

For what an okayish possible future could look like, I have two stories in mind:

1. Humans end up as housecats. Living among much more powerful creatures doing incomprehensible things, but still mostly cared for.

2. Some humans get uplifted to various levels, others stay baseline. The higher you go, the more aligned you must be to those below. So still a hierarchy, with super-smart creatures at the top and housecats at the bottom, but with more levels in between.

A post-AI world where baseline humans are anything more than housecats seems hard to imagine, I'm afraid. And even getting to be housecats at all (rather than dodos) looks to be really difficult.

Thanks for writing this, it's a great explanation-by-example of the entire housing crisis.

Well, Christianity sometimes spread by conquest, but other times it spread peacefully just as effectively. Same for democracy. So I don't think the spread of moral values requires conquest.

Wait, but we know that people sometimes have happy moments. Is the idea that such moments are always outweighed by suffering elsewhere? It seems more likely that increasing the proportion of happy moments is doable, an engineering problem. So basically I'd be very happy to see a world such as in the first half of your story, and don't think it would lead to the second half.

Your theory would predict that we'd be much better at modeling tigers (which hunted us) than at modeling antelopes (which we hunted), but in reality we're about equally bad at modeling either, and much better at modeling other humans.

I don't think this post addresses the main problem. Consider the exchange ratio between labor and land. You need land to live, and your food needs land to be grown. Will you be able to afford more land use for the same work hours, or less? (As programmer, manager, CEO, super high productivity job, whatever.) Well, if the same land can be used to run AIs that can do your job N times over, then from your labor you won't be able to afford it, and that closes the case.

So basically, the only way the masses can survive long term is by some kind of handouts. It won't just happen by itself due to tech progress and economic laws.

I don't buy it. Lots of species have predators and have had them for a long time, but very few species have intelligence. It seems more likely that most of our intelligence is due to sexual selection, a Fisherian runaway that accidentally focused on intelligence instead of brightly colored tails or something.

An ASI project would be highly distinguishable from civilian AI applications and not integrated with a state’s economy

Why? I think there's a smooth ramp from economically useful AI to superintelligence: AIs gradually become better at many tasks, and these tasks help more and more with improving AI in turn.

For cognitive enhancement, maybe we could have a system like "the smarter you are, the more aligned you must be to those less smart than you"? So enhancement would be available, but would make you less free in some ways.