What confuses me about the actual verification of these experiments is that they require perfect timing and distancing. How exactly do you make two photons hit a half-silvered mirror at exactly the same time? It seems that if you were off only slightly then the universe would necessarily have to keep track of both as individuals. In practice you’re always going to be off slightly so you would think by your explanation above that this would in fact change the result of the experiment, placing a 25% probability on all four cases. Why doesn’t it?
Jordan
So, I was thinking about the misapplication of the deduction theorem and suddenly had an insight into friendly AI. (Stop me if this has been discussed =D )
The problem is you give the AI a set of ‘axioms’ or goals and it might go off and tile the universe. Obviously we don’t know the full consequences of the logical system we’re initiating otherwise we wouldn’t need the AI in the first place.
The problem already exists in humans. Evolution gave us ‘axioms’ or desires. Some people DO go off and tile their universe: drugs, sex or work addictions, etc, etc. Thus my insight stems from the lack of drug addictions in most people. Here are my two proposed solutions:
(1) Fear. Many people don’t do drugs not because of a lack of desire to feel good, but because they are scared. People are likewise scared of any large changes (moving, new job, end of a relationship). Now, we don’t need the AI to favor status quo, however we can simply code into ones of its axioms that large physical changes are bad. Scale this exponentially (ie, twice the physical changes SQUARES the negative weighting of the action). Do not have any positive weighting criteria on other goals that scale faster than a polynomial.
(2) Life. Many people don’t tile their universe because they have too many different things they’d like to tile it with. Hobbies, friends, lovers, variation, etc. Give the AI a multitude of goals and set the positive weight associated with their accomplishment to diminish with returns (preferably logarithmic in growth at most). Twice the tiles only gives a linear increase in gauged utility.
Volume of tiling is bounded above by polynomial growth in time (cubic in a 3D universe with speed limit), hence hitting it with a log penalty will stifle its growth to at most log(t). If you wanted to be really safe you could simply cap the possible utility of accomplishing any particular goal.
I’m missing something because this seems like a solid solution to me. I haven’t read most of Eliezer’s writings, unfortunately (no time, I tile my universe with math), so if there’s a good one that discusses this I’d appreciate a link.
When I was a teenager I had a concept I referred to as “the double edged sword of apathy”. It was precisely the concept that separating oneself from certain aspects of oneself (which at the time I called fostering apathy) is a destructive tool which can be either positive or negative. Care must be taken not to slice your own arm off.
I don’t believe that this danger should be removed though, at least I wouldn’t personally allow it. I hold “self-modifying” to be the deepest aspect of life. When we finally get the technology to do source level modifications I won’t let an AI do the job: I’ll do it myself, regardless of the risk.
‘Fun’ is just a word. Your use of it probably doesn’t coincide with the standard meaning. The standard version of fun could likely be easily boxed in an orgasmium type state. You’ve chosen a reasonable sounding word to encapsulate the mechanisms of your own preference. Nietzsche would call that mechanism instinct, Crowley love. What it boils down to is that we all have a will, and that will is often counter to prior moralities and biological imperatives.
My own arbitrary word for preferable future states is ‘interesting’. You’d have to be me for that to really mean anything though.
In my mind this comes down to a fundamental question in the philosophy of math. Do we create theorems or discover them?
If it turns out to be ‘discovery’ then there is no foul in ending a mind emulation, because each consecutive state can be seen as a theorem in some formal system, and thus all states (the entire future time line of the mind) already exists, even if undiscovered.
Personally I fail to see how encoding something in physical matter makes the pattern anymore real. You can kill every mathematician and burn every text book but I would still say that the theorems then inaccessible to humanity still exist. I’m not so convinced of this fact that I would pull the plug on an emulation though.
Without pain can there be heroism?
Actually, this doesn’t sound like such a bad setup. Even the ‘catgirls’ wouldn’t be tiring, their exquisiteness intimately tied up in feelings of disgust and self-hate—probably a pretty potent concoction. The overarching quest to reunite with the other half of the species provides meaningful drive with difficult obstacles (science etc), but with a truly noble struggle baked within (the struggle against oneself).
I imagine a distant future with just a smattering of paper clip maximizers—having risen in different galaxies with slightly different notions of what a paperclip is—might actually be quite interesting. But even so, so what? Screw the paperclips, even if they turn out to be more elegant and interesting than us!
@Jotaf, “Order > chaos, no?”
Imagine God shows up tomorrow. “Everyone, hey, yeah. So I’ve got this other creation and they’re super moral. Man, moral freaks, let me tell you. Make Mennonites look Shintoist. And, sure, I like them better than you. It’s why I’m never around, sorry. Thing is, their planet is about to get eaten by a supernova. So.. I’m giving them the moral green light to invade Earth. It’s been real.”
I’d be the first to sign up for the resistance. Who cares about moral superiority? Are we more moral than a paperclip maximizer? Are human ideals ‘better’? Who cares? I don’t want an OfficeMax universe, so I’ll take up arms against a paperclip maximizer, whether its blessed by God or not.
By far the most enjoyable writing of yours I’ve read. I’d pay to read the rest but since you’re giving it up for free I’ll make a donation to SIAI instead.
Explaining science to laypeople with limited time.
Assuming agents participate in a sufficient number of public one-shot PD’s is essentially the same as playing iterated PD. If true one-shot PD’s are rare its doubtful there’d be enough historic evidence of your opponent to be certain of anything.
I’ve got a slew of digestion issues and some metabolic problems (first ulcer at age 13). Pertinent info I’ve learned:
1) Diarrhea is a hell of a way to lose weight.
2) Treating your diet like a controlled scientific experiment does wonders. For about a year I never at more than 2 − 4 ingredients per meal (an ingredient being a single, unprocessed, whole food). That was a tough year, and my diet remains restricted due to what I learned, but the health I’ve earned is invaluable. What helps me is to think of food as a source of fuel, not pleasure.
Not necessarily schools for all though. Uneducated people are easier to rule.
Also, a little bit of knowledge can be dangerous. Based on my observations of people with a little bit of political knowledge, I would guess slightly increasing the education of everyone across the board could prove disastrous in a democracy.
Two reasons.
One, in my experience people with slightly above average political knowledge tend to be more partisan. This could easily be a selection bias, partisans being more likely to learn a little, but I think in general whenever people learn they are more likely to use it to amplify their confidence in their current ideology.
Two, a little knowledge can create undue confidence. There are a lot of issues people (rightly) stay completely out of because they have zero knowledge. If everyone had just enough knowledge to feel engaged by an issue the decision makers might become pressured to follow the partially informed judgments of the majority.
I guess #1. Here are my five:
1) I’ve climbed Pão de Açucar
2) Every man I’ve kissed has been dressed like a woman
3) I’ve never had a single tooth cavity
4) I once hung off the wing of a plane (that was flying)
5) I’m going bald
I suppose I could clarify:
Every man I’ve kissed has been dressed in clothing traditionally accepted by Western culture to be feminine and moreover this group is nonempty.
Likewise by (3) I don’t mean that every time I’ve had a cavity I’ve had at least two.
Nice analysis. One comment though…
Ahzore sbhe vf vaqrrq cbffvoyr. V’ir qbar vg =Q.
Well, the plane was too small to jump out of directly without hitting the tail, so first I had to grab the wing and work my way to the tip before letting go. I’ve got to say, however much of a rush skydiving is, hanging from a wing is way more exhilarating.
The lie was (1), it’s definitely something I’d like to do, just haven’t got around to it.
At Constant, Is there a ‘natural’ probability measure on the set of all possible existences? Otherwise it has to be included in the ‘program’ and not the ‘RAM’.