Richard, am I reading you right when I'm interpreting you to be saying that there are several distinguished cogsci professionals who have read Yudkowsky's writings and have serious disagreements with his major points?
Well, you could take pretty much anything ever written about the theory of cognition, theoretical AI or philosophy of mind and find "several distinguished cogsci professionals who ... have serious disagreements with the major points." So it is hardly a surprise that this would be the case for Eliezer Yudkowsky's writings as well. Furthermore, Yudkowsky in year N often has serious disagreements with Yudkowsky in year (N-k) where k is a small integer ;-)
(Personally, I can't help but find Yudkowsky's writings consistently brilliant, so I'd quite curious in hearing reasons for why serious professionals would consider him a nutcase.
Brilliant is one thing; correct is another ;-) As for why some professionals would view Yudkowsky's ideas with skepticism, I suppose there are two reasons 1) disagreement with the ideas themselves [but as I noted above, professionals in cog sci and AI have wild disagreements with each other on foundational issues, all the time] 2) Yudkowsky's writings often are not written in the dry, reference-heavy style that academics favor Personally, I think Eliezer's writing style is wonderful, and that he has introduced a number of interesting ideas into the dialogue on AGI and the Singularity. I have disagreements with him on many essential points; but this is a contentious area where there is no consensus among experts. Of course, I think I am right in all our disagreements, though ;-) I do agree with Samantha and many others that "formulating a design for a pragmatically achievable, provably Friendly AI" (Eliezer's central goal these days) is almost surely an infeasible goal. However, I also think that the achieving of this goal would be SO NICE that it's well worth having some smart people spending some time working on it. I myself have spent some bits and pieces of time working on it, but have repeatedly given up in frustration and turned back to merely-very-hard goals like "creating an AGI according to a design that seems very likely to be Friendly." -- Ben G ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=4007604&user_secret=7d7fb4d8
