cubefox

Karma: 626

cubefox 23 Oct 2023 15:52 UTC
1 point
0
in reply to: Zane’s comment on: What is an “anti-Occamian prior”?
A hypothesis corresponds to an “event” in Kolmogorov probability theory. Such an event is the set of mutually exclusive and exhaustive “outcomes” under which the hypothesis is true. An outcome is basically a possible world. So a hypothesis can be thought of as the set of (or the disjunction of) possible worlds in which the hypothesis is true. So the probability of an event/hypothesis is exactly the sum of the probabilities of its outcomes.

That being said, this way of seeing things is problematic, since most hypotheses are true in infinitely many “possible worlds”. It is not clear how we could sum infinitely many probabilities. And if we don’t use possible worlds as outcomes, but use arbitrary DNFs, the decomposition (partition) becomes non-unique, because outcomes are no longer distinguishable from normal events. Possible worlds at least have the special feature of being “atomic” in the sense that they can’t be decomposed into mutually exclusive and exhaustive disjuncts.

Ideally we would want finitely many non-arbitrary outcomes for each hypothesis. Then we could apply the principle of indifference, assign equal probabilities, and create the sum. But that doesn’t seem to work.

So the whole disjunctive approach seems not exactly promising. An alternative is Wittgenstein-style logical atomism, which supposes there are something like “atomic propositions”, presumably about sense data, which are all independent of each other. Hypotheses/events are then supposed to be logical combinations of these. This approach is riddled with technical difficulties as well, it is unclear how it could be made precise.

cubefox 23 Oct 2023 10:51 UTC
2 points
0
on: What is an “anti-Occamian prior”?

It seems like Occam’s Razor just logically follows from the basic premises of probability theory. Assume the “complexity” of a hypothesis is how many bits it takes to specify under a particular method of specifying hypotheses, and that hypotheses can be of any length.

I think the method can’t be arbitrary for your argument to work. If we were to measure the “complexity” of an hypothesis by how many outcomes the event set of the hypothesis contains (or by how many disjuncts some non-full disjunctive normal form has), then the more “complex” hypotheses are less likely.

Of course that’s an implausible definition of complexity. For example, using conjunctive normal forms for measuring complexity is overall more intuitive, since $A \land B$ is considered more complex than $C$ , and it would indeed lead to more “complex” hypothesis being less likely. But universally quantified statements (like laws of nature) correspond to very long (perhaps infinitely long) CNFs, which would suggest that they are highly complex and unlikely. But intuitively, laws of nature can be simple and likely.

This was basically the problem Wittgenstein later identified with his definition of logical probability in the Tractatus a hundred years ago. Laws would have logical probability 0, which is absurd. It would mean no amount of finite evidence can confirm them to any degree. Later Rudolf Carnap worked on this problem, but I’m pretty sure it is still considered to be unsolved. If someone does solve it, and finds an a priori justification for Ockham’s razor, that could be used as a solution to the problem of induction, the core problem of epistemology.

So finding a plausible definition of hypothesis complexity, for which also Ockham’s razor holds, is a very hard open problem.

cubefox 22 Oct 2023 10:55 UTC
2 points
0
on: Fertility Roundup #2
I’m surprised this post doesn’t mention the arguably simplest and most obvious theory of fertility decline. I call it the female opportunity cost theory (FOC). It makes the following claims:
1. In relationships, the decision to have children is mostly influenced by women rather than men.
2. Whether a woman wants children depends on her perceived cost to have children.
3. Due to culture becoming more gender egalitarian, women are more and more expected to have careers rather than assuming the role of a housewife (while the husband is the breadwinner) in the traditional family model.
4. a) Women who already have careers rather than being housewives perceive a much higher opportunity cost of having children, since this means they miss out on (a substantial part of) their careers. A woman’s perceived opportunity cost is especially high when her male partner doesn’t have a substantially higher income than herself.
  b) Young women / teenage girls, who do not yet have careers or children, also perceive a high expected opportunity cost of becoming a mother rather than having a (bigger) career. This was different for girls who grew up in a more traditional society where it was normal for women to become a housewife and have children, while it was the job of the husband to earn money. The option of having a career wasn’t as much a live possibility for these young women as it is now, so they didn’t perceive not having a career as a price they had to pay for having children. Opportunity costs depend on social expectations, on what is considered the default.
This theory says that fertility decline is mainly caused by societies becoming more and more gender egalitarian, by women being expected to have careers rather than to become housewives.

The most important observation which supports this theory is that female IQ and education is negatively correlated with fertility. IQ and education cause better careers and career prospects, which (in a gender egalitarian society) increases the opportunity cost of having children instead.

(This also suggests the worrying consequence that not only overall fertility is decreasing, but also that fertility of more intelligent women is decreasing much faster than of less intelligent women, which would lead to an overall decrease in intelligence. A dysgenic trend.)

The reason why this theory apparently isn’t talked about much in academia is probably that it sounds sexist or ultraconservative. It makes it look like the achievements of woman’s liberation were a mistake. But things that may be good in one way can have bad consequences in another way. We might be inclined to wish the theory isn’t true, but wishful thinking never helps. We should instead wish to know the truth, because this is the only way to find out what we could do.

cubefox 19 Oct 2023 21:07 UTC
3 points
0
in reply to: jacob_cannell’s comment on: Evolution Solved Alignment (what sharp left turn?)

for example a dedicated jihadi who sacrifices their mortal life for reward in the afterlife is rather obviously pursing complex cultural (linguistically programmed) terminal values.

But that’s clearly an instrumental value. The expected utility of sacrificing his life may be high if he believes he will have an afterlife as a result. But if he just stops believing that, his beliefs (his epistemics) have changed, and so the expected utility calculation changes, and so the instrumental value of sacrificing his life has changed. His value for the afterlife doesn’t need to change at all.

You can imagine a jihadist who reads some epistemology and science books and as a result comes to believe that the statements in the Quaran weren’t those of God but rather of some ordinary human without any prophetic abilities. Which diminishes their credibility to those found in an arbitrary fiction book. So he may stop believing in God altogether. Even if only his beliefs have changed, it’s very unlikely he will continue to see any instrumental value in being a jihadist.

It’s like drinking a glass of clear liquid because you want to quench your thirst. Unfortunately the liquid contains poison, while you assumed it was pure water. Drinking poison here doesn’t mean you terminally value poison, it just means you had a false belief.

Regarding ethics/alignment: You talk about what people are actually motivated by. But this is arguably a hodgepodge of mostly provisional instrumental values that we would significantly modify if we changed our beliefs about the facts. As we have done in the past decades or centuries. First we may believe doing X is harmful, and so instrumentally disvalue X, then we may learn more about the consequences of doing X, such that we now believe X is beneficial or just harmless. Or the other way round. It would be no good to optimize an AI that is locked in with increasingly outdated instrumental values.

And an AI that always blindly follows our current instrumental values is also suboptimal: It may be much smarter than us, so it would avoid many epistemic mistakes we make. A child may value some amount of autonomy, but it also values being protected from bad choices by its smarter parents—choices which would harm it rather than benefit. An aligned ASI should act like such a parent: It should optimize for what we would want if we were better informed. (We may even terminally value some autonomy at the cost of making some avoidable instrumental mistakes, though only to some extent.) For this the AI has to find out what our terminal values are, and it has to use some method of aggregating them, since our values might conflict in some cases. That’s what normative ethics is about.

cubefox 19 Oct 2023 17:39 UTC
3 points
0
in reply to: jacob_cannell’s comment on: Evolution Solved Alignment (what sharp left turn?)
Hmm… Reading your last comment again, it seems you make some point similar to Robin Hanson’s Poor Folks Do Smile. An argument that evolution optimizes to some limited extent for happiness or satisfied preferences, to a degree significantly above neutral welfare lives. Add to that Hanson’s commitment to total (rather than average) utilitarianism, evolution is pretty aligned with ethics, as it optimizes for a high number of positive welfare individuals. I’m not sure how his argument is supposed to generalize to a world with AI though. Presumably we humans don’t want to be replaced with even fitter AI, while evolutionary pressures arguably do “want” this. And total utilitarianism (the Repugnant Conclusion) seems unreasonable to me anyway.

But apart from your similarity to Hanson: I think the goal of AI alignment really is ethics. Not some specific human values shaped by millennia of biological or even cultural evolution. First, I don’t think cultural evolution influences our terminal values. And while biological evolution does indeed shape terminal values, it operates fundamentally on the level of individual agents. But ethics is about aggregate welfare. A society of Martians may have different terminal values from us (they may not fear death after procreation, for example), but they could still come up with the same forms of utilitarianism as we do. Ethics, at least if some sort of utilitarianism is right, isn’t about labelling some idiosyncratic collection of terminal values as “good” that we historically evolved to have, but rather about maximizing an aggregate (like average or sum) of terminal values (happiness or desire satisfaction) of individuas, whatever those individual values may be. I think biological evolution doesn’t very strongly optimize for the total aggregate (though Hanson may disagree), and hardly at all for the average.

Though admittedly I don’t fully understand your position here.

cubefox 19 Oct 2023 12:57 UTC
3 points
2
in reply to: jacob_cannell’s comment on: Evolution Solved Alignment (what sharp left turn?)
My point is that organisms become, over generations, aligned to IGF automatically, simply because organisms who have more offspring outcompete others in the future. But no such automatism exists for purposefully engineering systems that optimize for other aims. Which does strongly suggest it is more likely to go wrong.

cubefox 19 Oct 2023 12:32 UTC
1 point
0
in reply to: abramdemski’s comment on: Translations Should Invert
The point of axiomatizing aspects of natural language reasoning, like decision theory, is to make them explicit, systematic, and easier to reason about. But the gold standard remain things that are valid in our antecedent natural language understanding. The primitive terms of any axiomatic theory are only meaningful insofar they reflect the meaning of some natural language terms, and the plausibility of the axioms derives from those natural language interpretations.

So for example, when we compare the axiomatizations of Savage and Jeffrey, we can do so by comparing how well or to what extent they capture the reasoning that is plausible in natural language. I would argue that Jeffrey’s theory is much more general, it captures parts of natural language reasoning that couldn’t be expressed in Savage’s earlier theory, while the opposite is arguably not the case. We can argue about that in English, e.g. by using terms like “action” with its natural language meaning and by discussing which theory captures it better. Savage assumes that outcomes are independent of “actions”, which is not presumed when doing practical reasoning expressed in natural language, and Jeffrey captures this correctly. One could object that Jeffrey allows us to assign probabilities to our own actions, which might be implausible, etc.

cubefox 15 Oct 2023 8:10 UTC
1 point
0
on: Noticing confusion in physics

If you increase the temperature, density can increase, but pressure also increases by a portional amount since the molecules are hitting harder. These two changes cancel out.

I think you meant to say “density can decrease” not “increase”?

cubefox 14 Oct 2023 22:26 UTC
1 point
in reply to: Chris_Leong’s comment on: Chris_Leong’s Shortform
I argue that “I can do X” means “If I want to do X, I will do X”. This can be true (as an unqualified statement) even with determinism. It is different from saying that X is physically possible.

cubefox 14 Oct 2023 11:21 UTC
3 points
1
in reply to: RobertM’s comment on: Evolution Solved Alignment (what sharp left turn?)
I don’t think it looks like we are about to achieve escape velocity. The broader laws of natural selection and survival of the fittest even apply to future AI and Grabby Aliens. Essentially the argument from “Meditations On Moloch”.

cubefox 12 Oct 2023 18:34 UTC
3 points
2
in reply to: avturchin’s comment on: Evolution Solved Alignment (what sharp left turn?)
But natural selection favors people who prefer having children, and various forms of wireheading eventually stop working if they decrease fitness.

cubefox 12 Oct 2023 18:22 UTC
12 points
11
in reply to: AnthonyC’s comment on: Evolution Solved Alignment (what sharp left turn?)
Subpopulations which do this are expected to disappear relatively quickly in evolutionary time scales. Natural selection is error correcting. This can mean people get less intelligent again, or they start to really love getting children rather than enjoying sex.

cubefox 12 Oct 2023 18:10 UTC
3 points
2
in reply to: TekhneMakre’s comment on: Evolution Solved Alignment (what sharp left turn?)
Good question. I think a good approximation is total body mass of the population. By this measure, only cattle are more successful than humans, and obviously only because we deliberately breed them for food: https://xkcd.com/1338/

cubefox 12 Oct 2023 11:06 UTC
4 points
2
on: Evolution Solved Alignment (what sharp left turn?)
I agree that the optimization process of natural selection likely irons out any cases of temporary misalignment (some human populations having low fertility) over a medium time span. People who tend to not have children eventually get replaced by people who love children, or people who tend to forget to use contraceptives etc. This is basically the force Scott Alexander calls Moloch.

Unfortunately this point doesn’t obviously generalize to AI alignment. Natural selection is a simple, natural optimization process, which optimizes a simple “goal”. But getting an AI to align with human ethics might be much harder, because human ethics might be a target much easier to miss, and not one that nature automatically “aims” for by itself, as it does for IGF.

cubefox 12 Oct 2023 10:51 UTC
4 points
8
in reply to: TekhneMakre’s comment on: Evolution Solved Alignment (what sharp left turn?)

Nitpick arguments about how you define this specifically are irrelevant and uninteresting.

Excuse me, what? This is not evolution’s utility function. It’s not optimizing for gene count. It does one thing, one thing only, and it does it well: it promotes genes that increase their RELATIVE FREQUENCY in the reproducing population.

I don’t see how this detail is relevant. The fact remains that humans are, in evolutionary terms, much more successful than most other mammals.

The failure of alignment is witnessed by the fact that humans very very obviously fail to maximize the relative frequency of their genes in the next generation, given the opportunities available to them; and they are often aware of this; and they often choose to do so anyway. The whole argument in this post is totally invalid.

Currently the world fertility rate is 2.3 children per woman. This is more than the often quoted 2.1 for a stable population. It is true that many developed countries are currently below that value. But this only means that these subpopulations will go extinct, while other subpopulations with higher fertility inherit the world. E.g. in Africa, Muslim countries, the Amish etc.

For example, in women high fertility is associated with, and likely caused by, low IQ and low education. The popular theory is that smart well-educated women have profitable and satisfying careers and consequently see much higher opportunity cost for having children instead, so they have fewer or none. And people with cognitive properties that cause them to have more children likely are currently getting more frequent in the population. Same as men who tend to forget to use condoms etc. These outlooks are not rosy from an ethical perspective, since properties like low intelligence and low conscientiousness are associated with a lot of human hardship such as poverty and violence, but they are totally in line with what evolution optimizes for.

cubefox 11 Oct 2023 7:39 UTC
3 points
−2
on: Translations Should Invert
The way out of such translation problems is usually natural language. In the logic case, both interlocutors presumably speak the same language, so they can both use the “logic of English” (of what follows from what in English natural language statements). Unlike some formal logic, the logic of English may be more vague, perhaps even probabilistic instead of deterministic, more evidence conferring than truth preserving, but these things are there for both interlocutors equally, so no translation is necessary, and things can be clarified by using statements about English statements made in English itself. If there is really a disagreement, it can be reduced to how some simple sentence in natural language should be interpreted in some hypothetical situation (thought experiment). These language intuitions are all over philosophy.

I’m not sure where the underlying disagreement in the decision theory case was (something about actions vs mixed strategies) but I assume there again the underlying problem can be expressed in natural language statements which both parties can understand without the need of translating them.

cubefox 5 Oct 2023 18:39 UTC
1 point
0
in reply to: Bezzi’s comment on: Dall-E 3
Yeah, though Dall-E 3 generally has better language understanding than other text-to-image models. (See e.g. here) I still think the “experimental” approach is more interesting for me personally than the deliberate one you describe. For example, with the previous Bing Image Creator (Dall-E 2.5), I “explored” photographs of fictional places, like Tlön, Uqbar, and strange art in an abandoned museum in Atlantis. It is a process of discovery rather than targeted creation. It’s probably personal preference. I’m not very creative, so I wouldn’t know what to draw if I could draw.

cubefox 5 Oct 2023 6:57 UTC
1 point
0
in reply to: Portia’s comment on: Dall-E 3
These models are still less for making you what you already have in mind than for trying out creative and outlandish prompts, varying and combining them, being occasionally surprised by a picture looking really great, and getting a feel for the “talents” of the model.

cubefox 22 Sep 2023 11:24 UTC
−1 points
0
in reply to: Ape in the coat’s comment on: Reflexive decision theory is an unsolved problem
Yes, but I was taking about humans. An AI might have a precommitment ability.

cubefox 22 Sep 2023 6:03 UTC
2 points
1
on: The Talk: a brief explanation of sexual dimorphism
Suggestion: Maybe move the summary to the top (as a sort of abstract or tl;dr).