dxu

dxu's Posts

Sorted by New

dxu's Comments

Predictors exist: CDT going bonkers... forever

these examples can't actually happen, or are so rare that I'll pay that cost in order to have a simpler model for the other 99.9999% of my decisions

Indeed, if it were true that Newcomb-like situations (or more generally, situations where other agents condition their behavior on predictions of your behavior) do not occur with any appreciable frequency, there would be much less interest in creating a decision theory that addresses such situations.

But far from constituting a mere 0.0001% of possible situations (or some other, similarly minuscule percentage), Newcomb-like situations are simply the norm! Even in everyday human life, we frequently encounter other people and base our decisions off what we expect them to do—indeed, the ability to model others and act based on those models is integral to functioning as part of any social group or community. And it should be noted that humans do not behave as causal decision theory predicts they ought to—we do not betray each other in one-shot prisoner’s dilemmas, we pay people we hire (sometimes) well in advance of them completing their job, etc.

This is not mere “irrationality”; otherwise, there would have been no reason for us to develop these kinds of pro-social instincts in the first place. The observation that CDT is inadequate is fundamentally a combination of (a) the fact that it does not accurately predict certain decisions we make, and (b) the claim that the decisions we make are in some sense correct rather than incorrect—and if CDT disagrees, then so much the worse for CDT. (Specifically, the sense in which our decisions are correct—and CDT is not—is that our decisions result in more expected utility in the long run.)

All it takes for CDT to fail is the presence of predictors. These predictors don’t have to be Omega-style superintelligences—even moderately accurate predictors who perform significantly (but not ridiculously) above random chance can create Newcomb-like elements with which CDT is incapable of coping. I really don’t see any justification at all for the idea that these situations somehow constitute a superminority of possible situations, or (worse yet) that they somehow “cannot” happen. Such a claim seems to be missing the forest for the trees: you don’t need perfect predictors to have these problems show up; the problems show up anyway. The only purpose of using Omega-style perfect predictors is to make our thought experiments clearer (by making things more extreme), but they are by no means necessary.

Realism about rationality

That depends on how strict your criteria are for evaluating “similarity”. Often concepts that intuitively evoke a similar “feel” can differ in important ways, or even fail to be talking about the same type of thing, much less the same thing.

In any case, how do you feel law thinking (as characterized by Eliezer) relates to the momentum-fitness distinction (as characterized by ricraz)? It may turn out that those two concepts are in fact linked, but in such a case it would nonetheless be helpful to make the linking explicit.

Realism about rationality

Doesn't the law thinker position imply that intelligence can be characterized in a "lawful" way like momentum?

It depends on what you mean by "lawful". Right now, the word "lawful" in that sentence is ill-defined, in much the same way as the purported distinction between momentum and fitness. Moreover, most interpretations of the word I can think of describe concepts like reproductive fitness about as well as they do concepts like momentum, so it's not clear to me why "law thinking" is relevant in the first place--it seems as though it simply muddies the discussion by introducing additional concepts.

Debate on Instrumental Convergence between LeCun, Russell, Bengio, Zador, and More

Skimming through. May or may not post an in-depth comment later, but for the time being, this stood out to me:

I think it would only be relevant in a fantasy world in which people would be smart enough to design super-intelligent machines, yet ridiculously stupid to the point of giving it moronic objectives with no safeguards.

I note that Yann has not actually specified a way of not "giving [the AI] moronic objectives with no safeguards". The argument of AI risk advocates is precisely that the thing in quotes in the previous sentence is difficult to do, and that people do not have to be "ridiculously stupid" to fail at it--as evidenced by the fact that no one has actually come up with a concrete way of doing it yet. It doesn't look to me like Yann addressed this point anywhere; he seems to be under the impression that repeating his assertion more emphatically (obviously, when we actually get around to building the AI, we'll use our common sense and build it right) somehow constitutes an argument in favor of said assertion. This seems to be an unusually low-quality line of argument from someone who, from what I've seen, is normally much more clear-headed than this.

What explanatory power does Kahneman's System 2 possess?

I'm curious as to what prompted this question?

Weak foundation of determinism analysis

I have pointed out what people worry they are going to lose under determinism. Yes, they only going to have those things under nondeterminism.

You just said that nondeterminist intuitions are only mistaken if determinism is true and compatibilism is false. So what exactly is being lost if you subscribe to both determinism and compatibilism?

Weak foundation of determinism analysis

You're mixing levels. If someone can alter their decisions, that implies there are multiple possible next states of the universe

This is incorrect. It's possible to imagine a counterfactual state in which the person in question differs from their actual self in an unspecified manner, which thereby causes them to make a different decision; this counterfactual state differs from reality, but it is by no means incoherent. Furthermore, the comparison of various counterfactual futures of this type is how decision-making works; it is an abstraction used for the purpose of computation, not something ontologically fundamental to the way the universe works--and the fact that some people insist it be the latter is the source of much confusion. This is what I meant when I wrote:

Decision-making itself is also a process that occurs in the map, not the territory; there is no contradiction here.

So there is no "mixing levels" going on here, as you can see; rather, I am specifically making sure to keep the levels apart, by not tying the mental process of imagining and assessing various potential outcomes to the physical question of whether there are actually multiple physical outcomes. In fact, the one who is mixing levels is you, since you seem to be assuming for some reason that the mental process in question somehow imposes itself onto the laws of physics.

(Here is a thought experiment: I think you will agree that a chess program, if given a chess position and run for a prespecified number of steps, will output a particular move for that position. Do you believe that this fact prevents the chess program from considering other possible moves it might make in the position? If so, how do you explain the fact that the chess program explicitly contains a game tree with multiple branches, the vast majority of which will not in fact occur?)

There are various posts in the sequences that directly address this confusion; I suggest either reading them or re-reading them, depending on whether you have already.

Weak foundation of determinism analysis

But this is map, not territory.

Certainly. Decision-making itself is also a process that occurs in the map, not the territory; there is no contradiction here. Some people may find the idea of decision-making being anything but a fundamental, ontologically primitive process somehow unsatisfying, or even disturbing, but I submit that this is a problem with their intuitions, not with the underlying viewpoint.

(If someone goes so far as to alter their decisions based on their belief in determinism--say, by lounging on the couch watching TV all day rather than being productive, because their doing so was "predetermined"--I would say that they are failing to utilize their brain's decision-making apparatus. (Or rather, that they are not using it very well.) This has nothing to do with free will, determinism, or anything of the like; it is simply a (causal) consequence of the fact that they have misinterpreted what it means to be an agent in a deterministic universe.)

Weak foundation of determinism analysis

Belief in determinism is correlated with worse outcomes, but one doesn't cause the other; both are determined by the state and process of the universe.

Read literally, you seem to be suggesting that a deterministic universe doesn't have cause and effect, only correlation. But this reading seems prima facie absurd, unless you're using a very non-standard notion of "cause and effect". Are you arguing, for example, that it's impossible to draw a directed acyclic graph in order to model events in a deterministic universe? If not, what are you arguing?

Drive-By Low-Effort Criticism

We didn’t (or rather, shouldn’t) intend to reward or punish those “ancestor nodes”. We should intend to reward or punish the results.

I'm afraid this sentence doesn't parse for me. You seem to be speaking of "results" as something which to which the concept of rewards and punishments are applicable. However, I'm not aware of any context in which this is a meaningful (rather than nonsensical) thing to say. All theories of behavior I've encountered that make mention of the concept of rewards and punishments (e.g. operant conditioning) refer to them as a means of influencing behavior. If there's something else you're referring to when you say "reward or punish the results", I would appreciate it if you clarified what exactly that thing is.

Load More