Re: Maximizing vs proving friendliness

From: Stuart Armstrong (
Date: Fri May 02 2008 - 06:13:03 MDT

> >You cannot have any help from AI because that would mean the AI is
> >helping to reprogram its own goals.
> You're splitting things into two pieces when you don't need to, and
> then arguing that each piece must precede the other so it can't be
> done. The two pieces are the AI writing code and the AI determining
> what its goals are. It is possible to solve both problems at once by
> putting the code generation into the utility function.

Also, I see no problem in having an AI dedicated to doing nothing but
working internally on a particular task, printing out the results two
days later, then turning itself off. We can use limited AI's to help
design general AI's.


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT