Comment comments on A cynical explanation for why rationalists worry about FAI - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (179)
Grandiosity, belief in own special importance, etc.
Narcissists are pretty common, people capable of grand contributions are very rare, so the majority of people who think they are capable of grand contributions got to be narcissists.
Speaking of which, Yudkowsky making a friendly AI? Are you frigging kidding me? I came here through the link to guy's quantum ramblings, which are anything but friendly.
Eliezer argues that a lot of people are more capable than they permit themselves to be, which doesn't seem very narcissistic to me.
From your link:
Nope. Crackpots compare themselves to Einstein because:
[Albeit I do like his straight in your face honesty.]
It's not about choosing 'important' problem, it's about choosing solvable important problem, and a method of solving it, and intelligence helps, while unintelligent people just pick some idea out of science fiction or something, and can't imagine that some people can do better.
Had it really been that choosing right problems and approaches was matter of luck, we would observe far fewer cases where a single individual has many important insights, the distribution of insights per person would be different.
edit:
The irony here is quite intense. Surely a person who's into science fiction will have the first "cache hit" be something science fictional, and then the first "cache hit" for the solution path be something likewise science fictional. Also a person into reading about computers will have first "cache hit" for describing the priming be reference to "cache".
Richard Hamming also makes this point.
Thanks. He says it much more better than I could. He speaks of importance of small problems.
Speaking of which, one thing geniuses do is generate the right problems for themselves, not just choose from already formulated.
Science fiction is full of artificial minds, good and evil. It has minds improving themselves, and a plenty of Frankensteins of all kinds. It doesn't have things like 'a very efficient universal algorithm that given mathematical description of a system and constraints finds values for free parameters that meet constraints', because it is not a plot device. Fiction does not have wolfram alpha in 2010. It has Hal in 2000 . Fiction shuns merely useful in favor of interesting. I would be very surprised if the solution would be among the fictional set. The fictional set is as good place to look in as any, yes, but it is small. edit: On second thought, what I mean is that it would be very bad to be either inspired or 'de-spired' by fiction to any significant extent.
Yes: Humans think in stories but there are far far more concepts that do not make good story than do make it.