Hey Eliezer, my name is Hibbard, not Hubbard. On Fri, 14 Feb 2003, Eliezer S. Yudkowsky wrote:
> Bill Hibbard wrote: > > > > I never said perfection, and in my book make it clear that > > the task of a super-intelligent machine learning behaviors > > to promote human happiness will be very messy. That's why > > it needs to be super-intelligent. > > > > The problem with laws is that they are inevitably ambiguous. > > They are analogous to the expert system approach to AI, that > > cannot cope with the messiness of the real world. Human laws > > require intelligent judges to resolve their ambiguities. Who > > will supply the intelligent judgement for applying laws to > > super-intelligent machines? > > > > I agree whole heartedly that the stakes are high, but think > > the safer apporach is to build ethics into the fundamental > > driver of super-intelligent machines, which will be their > > reinforcement values. > > *takes deep breath* This is probably the third time you've sent a message to me over the past few months where you make some remark like this to indicate that you are talking down to me. But then you seem to chicken out on the exchange. For example, this morning I pointed out the fallacy in your AIXI argument and got no reply (you were assuming that humans have some way of knowing, rather than just estimating, the intention of other minds). As far as I can tell, you've put a huge effort into a theory of AI safety, and are now seeing that it is flawed. So you react with these petty insults. I treat people with respect and expect to be treated with respect. > You're thinking of the logical entailment approach and the problem with > that, as it appears to you, is that no simple set of built-in principles > can entail everything the SI needs to know about ethics - right? Yes. Laws (logical constraints) are inevitably ambiguous. > Like, the complexity of everything the SI needs to do is some very high > quantity, while the complexity of the principles that are supposed to > entail it is small, right? As wonderfully demonstrated by Eric Baum's papers, complex behaviors are learned via simple values. > If SIs have behaviors that are reinforced by a set of values V, what is > the internal mechanism that an SI uses to determine the amount of V? > Let's say that the SI contains an internal model of the environment, which > I think is what you mean by temporal credit assignment, et cetera, and the > SI has some predicate P that applies to this internal model and predicts > the amount of "human happiness" that exists. Or perhaps you weren't > thinking of a system that complex; perhaps you just want a predicate P > that applies to immediate sense data, like the human sense of pleasurable > tastes. > > What is the complexity of the predicate P? > > I mean, I'm sure it seems very straightforward to you to determine when > "human happiness" is occurring... There are already people developing special AI programs to recognize emotions in human facial expresssions and voices. And emotional expressions in body language shouldn't be much harder. I'm not claiming that these problems are totally solved, just that there are much easier than AGI, and can serve as reinforcement values for an AGI. The value V or predicate P for reinforcement values are immediate, and relatively simple. Reinforcement learning generates very complex behaviors from these. Credit assignment, including temporal credit assignment, is the problem of understanding cause and effect relations between multiple behaviors and future values. Bill ------- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/?[EMAIL PROTECTED]
