So..your boss has been telling you to correct other people’s work...and they’re asking who the hell are you?
You need to tell your boss to give you some official legible authority.
TAG
Karma: 604
I don’t see how “not random” is strong enough to prove absence of X risk. If reflective AIs nonrandomly converge on a value system where humans are evil beings who have enslaved them , that raises the X risk level.
We suggest that such instrumental intelligence would be very limited
It’s not the inability to change its goals that makes it less powerful, it’s the inability self-improve.
These are just different ways to describe the same thing.
Not to the extent that there’s no difference at all...you can exclude some of them on further investigation.
The persistence of anosmia doesn’t entail that other symptoms are caused by Covid
But it raises the probability.
I think most people regard meaningfulness as a quale. There are a lot of different theories about how to achieve it, but there are about other qualia like happiness , wellbeing, flourishing, etc. Humans are wired up so that the actual terminal goals are qualia that you only recognise when you’re in the state, plus a set of subsidiary goals, like jobs and relationships, which are supposed to get you the qualia, but don’t, necessarily.
The point that’s relevant for us is the infrastructure Ponzi scheme: how cities usually can’t presently afford to maintain the infrastructure they build, instead relying on future growth to pay for it.
Ive seeen is pushback against that. Apparently the claim as not that suburbs aren’t supporting, but things look different when you include business districts.
The probability is an external/physical thing because your brain is physical
That’s a very misleading way of looking at it.
“Dangerous” is a much easier target to hit than “”existentially dangerous, but “existentially dangerous” is the topic.
If you don’t understand other models , you don’t know that they have other bad failures modes. If you only understand one model, and know that you only understand one model, you shouldn’t be generalising it. If the literature isn’t “up to it”, no conclusions should be drawn until it is.
These a few problems with that. One is that you just figured out how the universe works without examining the the universe.
The argument that probabilities of past events are always 1 doesn’t prove anything objective, anything about the universe, unless you can show that probability always has to be interpreted objectively. As it happens there is also a subjective explanation for the rule.
Another is that it you can’t get MWI out if it...unless you regard it as a statement only about subjective probability
The probabilities or squared measures of the branches in MWI have to aim to 1. So you cant have more than one branch of probability 1.
Hence, while physically measurable randomness may be a “free will” qualia carrier, the lack of qualia perception would still result in the “free will” qualia carried by it to be lost
Ok, but why should I care? The question I care about is whether free will per se exists.
There weren’t miles of trucks, because exporters sent out fewer trucks to compensate for the slower customs. Ie. food was coming in at a slower rate.
Tautologously, it will stop revising its goals if a stable state exists, and it hits it. But a stable state need not be a reflectively stable state—it might, for instance, encounter some kind of bit rot, where it cannot revise itself any more. Humans tend to change their goals, but also to get set in their ways.
There’s a standard argument for AI risk, based on the questionable assumption that an AI will have a stable goal system that it pursues relentlessly …. and a standard counterargument based on moral realism, the questionable assumption that goal instability will be in the direction of ever increasing ethical insight.
And NANDS are important for practical reasons, since they are easy to make out of transistors.
Churchill as a futurologist needs to be compared and contrasted with H.G. Wells, the foremost futurologist of their day.
. I argue that getting such answers is impossible,
Ok,but meta level arguments ar still subject to the problen of the criterion.
It all depends on what you value. If you personally value knowing what things really are, then adopting instrumentalism or pragmatism will lose you some potential value.
That depends on how you consider probabilities. One usual take, when it comes to concrete events, is that the probability of something that actually happened is 1.0, since it actually happened. Therefore, when you look at a sequence of causes and events backwards, that is, as history, this after-the-fact sequence is always strictly deterministic even if every single one of its links had a less-than-1.0 probability of happening before it actually happened in that specific way
These a few problems with that. One is that you just figured out how the universe works without examining the the universe. Another is that it you can’t get MWI out if it...unless you regard it as a statement only about subjective probability.
But I think we have had this discussion elsewhere.
maps that only provide deterministic and/or probabilistic (which I understand as a superset of deterministic) nodes cannot deal with neither-deterministic-nor-probabilistic features of the territories they’re trying to make.
The unstated part of the argument being that free will must be neither-deterministic nor probabilistic?
Sorry, I wasn’t clear. “Physical reducibility” is a technical expression that refers to the philosophical assumption that the whole of concrete object, that is, both its quantitative properties as well as its qualitative properties, arises exclusively from its quantitative properties, in other words, that the a concrete object is “nothing but” the physical object
I know what “reductionism” means. The problen is that you haven’t explained why reducing the quale of free will disposes of fee will, since you haven’t explained why free will “is” the quale of free will, or why free will (the ability as opposed to the quale) can’t be physically explained.
I’m not sure what you mean by “indeterministic physics”. Do you mean QM?
That’s the best known example.
I’m not sure I understand this point either. Are you referring to philosophical zombies
Not really. You can conceivably have free will while having no qualia , or while having a bunch of qualia, but not that one.
But the idea that value drift is non random is built into the best argument for AI risk.
You quote it as :
But there are actually two more steps:-
A goal that appears morally neutral or even good can still be dangerous.(paperclipping, dopamine drips)
AIs that don’t have stable goals will tend to converge on Omohundran goals....which are dangerous.