Wednesday, September 22, 2004

Interesting stuff, I met Michael Roy Ames yesterday, He's starting a canadian branch of the Singularity Institute.

He's a commercial coder who is now in the field of AI as an activist and enthusiast. He began implementing an AI some time ago, but stopped because of the lack of specification in goal system, and because of the friendliness problem.

How does one ensure that AIs grow up to be nice people who do nice things? Or even, do nice things, if your AI is not a proper person.

No comments: