outlawpoet

Posts

Sorted by New

Comments

Advice for AI makers

That is an excellent question.

Call for new SIAI Visiting Fellows, on a rolling basis

I must agree with this, although video and most writing OTHER than short essays and polemics would be mostly novel, and interesting.

A Nightmare for Eliezer

If every snide, unhelpful jokey reply you post is secretly a knowing reference to something only one other person in the world can recognize, I retract every bad thing I ever said about you.

Privileging the Hypothesis

Is then the ability to explicitly (at a high, abstract level) reach down to the initial hypothesis generation and include, raise, or add hypotheses for consideration always a pathology?

I can imagine a system where extremely low probability hypotheses, by virtue of complexity or special evidence required, might need to be formulated or added by high level processes, but you could simply view that as another failure of the generation system, and require that even extremely rare or novel structures of hypotheses must go through channels to avoid this kind of disturbance of natural frequencies, as it were.

Privileging the Hypothesis

It may not be a completely generic bias or fallacy, but it certainly can affect more than just human decision processes. There are a number of primitive systems that exhibit pathologies similar to what Eliezer is describing, speech recognition systems, for example, have a huge issue almost exactly isomorphic to this. Once some interpretation of a audio wave is a hypothesis, it is chosen in great excess to it's real probability or confidence. This is the primary weakness of rule-based voice grammars, that their pre-determined possible interpretations lead to unexpected inputs being slotted into the nearest pre-existing hypothesis, rather than leading to a novel interpretation. The use of statistical grammars to try to pound interpretations to their 'natural' probabilistic initial weight is an attempt to avoid this issue.

This problem is also hidden in a great many AI decision systems within the 'hypothesis generation' system, or equivalent. However elegant the ranking and updating system, if your initial possible list is weak, you distort your whole decisions process.

The Finale of the Ultimate Meta Mega Crossover

The bloodstained sweater in the original song refers to an urban legend that Mr. Rogers was a Marine Sniper in real life.

Utilons vs. Hedons

Why on earth wouldn't I consider whether or not I would play again? Am I barred from doing so?

If I know that the card game will continue to be available, and that Omega can truly double my expected utility every draw, either it's a relatively insignificant increase of expected utility over the next few minutes it takes me to die, in which case it's a foolish bet, compared to my expected utility over the decades I have left, conservatively, or Omega can somehow change the whole world in the radical fashion needed for my expected utility over the next few minutes it takes me to die to dwarf my expected utility right now.

This paradox seems to depend on the idea that the card game is somehow excepted from the 90% likely doubling of expected utility. As I mentioned before, my expected utility certainly includes the decisions I'm likely to make, and it's easy to see that continuing to draw cards will result in my death. So, it depends on what you mean. If it's just doubling expected utility over my expected life IF I don't die in the card game, then it's a foolish decision to draw the first or any number of cards. If it's doubling expected utility in all cases, then I draw cards until I die, happily forcing Omega to make verifiable changes to the universe and myself.

Now, there are terms at which I would take the one round, IF you don't die in the card game version of the gamble, but it would probably depend on how it's implemented. I don't have a way of accessing my utility function directly, and my ability to appreciate maximizing it is indirect at best. So I would be very concerned about the way Omega plans to double my expected utility, and how I'm meant to experience it.

In practice, of course, any possible doubt that it's not Omega giving you this gamble far outweighs any possibility of such lofty returns, but the thought experiment has some interesting complexities.

Utilons vs. Hedons

I see, I misparsed the terms of the argument, I thought it was doubling my current utilons, you're positing I have a 90% chance of doubling my currently expected utility over my entire life.

The reason I bring up the terms in my utility function, is that they reference concrete objects, people, time passing, and so on. So, measuring expected utility, for me, involves projecting the course of the world, and my place in it.

So, assuming I follow the suggested course of action, and keep drawing cards until I die, to fulfill the terms, Omega must either give me all the utilons before I die, or somehow compress the things I value into something that can be achieved in between drawing cards as fast as I can. This either involves massive changes to reality, which I can verify instantly, or some sort of orthogonal life I get to lead while simultaneously drawing cards, so I guess that's fine.

Otherwise, given the certainty that I will die essentially immediately, I certainly don't recognize that I'm getting a 90% chance of doubled expected utility, as my expectations certainly include whether or not I will draw a card.

Utilons vs. Hedons

I seem to have missed some context for this, I understand that once you've gone down the road of drawing the cards, you have no decision-theoretic reason to stop, but why would I ever draw the first card?

A mere doubling of my current utilons measured against a 10% chance of eliminating all possible future utilons is a sucker's bet. I haven't even hit a third of my expected lifespan given current technology, and my rate of utilon acquisition has been accelerating. Quite aside from the fact that I'm certain my utility function includes terms regarding living a long time, and experiencing certain anticipated future events.

Load More