Reviewers wanted for New Book—When Computers Can Really Think.
The book aims at a general audience, and does not simply assume that an AGI can be built. It differs from others by considering how natural selection would ultimately shape a AGI’s motivations. It argues against the Orthogonality Principal, suggesting instead that there is ultimately only one super goal, namely the need to exist. It also contains a semi-technical overview of artificial intelligent technologies for the non-expert/student.
It argues against the conjecture that utility function is separate from optimization power? Do you mean that it argues against Omohundro’s instrumental AI drives?
Yes, I know; they’re convergent. I’m questioning what aberglas is arguing against with his Darwinist supergoal. It doesn’t make sense to say that such a supergoal is mutually exclusive with the independence of utility and optimization power. It makes more sense to say that the supergoal is an alternative to Omohundro’s instrumental drives.
I don’t see how what aberglas wrote makes coherent sense.
Reviewers wanted for New Book—When Computers Can Really Think.
The book aims at a general audience, and does not simply assume that an AGI can be built. It differs from others by considering how natural selection would ultimately shape a AGI’s motivations. It argues against the Orthogonality Principal, suggesting instead that there is ultimately only one super goal, namely the need to exist. It also contains a semi-technical overview of artificial intelligent technologies for the non-expert/student.
An overview can be found at
www.ComputersThink.com
Please let me know if you would be interested in reviewing a late draft. Any feedback would be most welcome. Anthony@berglas.org
I’m totally down, cameron@cameroncowan.net
I’m always happy to proofread. PM me with the details.
It argues against the conjecture that utility function is separate from optimization power? Do you mean that it argues against Omohundro’s instrumental AI drives?
The whole point of instrumental drives is that they don’t have to be in the utility function.
Yes, I know; they’re convergent. I’m questioning what aberglas is arguing against with his Darwinist supergoal. It doesn’t make sense to say that such a supergoal is mutually exclusive with the independence of utility and optimization power. It makes more sense to say that the supergoal is an alternative to Omohundro’s instrumental drives.
I don’t see how what aberglas wrote makes coherent sense.
Well, alternative if you like. I will post an elaboration as a full article.