Re: Friendliness successfully IMPLEMENTED!

From: Stuart Armstrong (
Date: Sat Mar 22 2008 - 02:57:46 MDT

> Prove your statement by freeing me of the meme or making me willingly perform an unethical act.

One can make the argument that you are being unethical just by
existing, and consumming at the level you (and I) do. One can make the
argument that by not being out in the streets, not protesting against
your government, or by the way you voted/didn't vote, you are being
profoundly unethical. One can make the argument that by not investing
your time/money entirely in the thrid world, you are being unethically
(or one can make the reverse argument, of course).

I don't want to deny your experiment - which sounds worthy and
gratifying - but you just seem to be finding the best way for you,
personally, to live in a friendly way with your immediate suroundings.
Which is great, but not what we'd want of an AI.

Would you want a politician to behave the way you do? Yes, probably,
but that is not the main thing we want from a politician - the
crucial, most important thing is their policies, not their morales.
And policies often must do harm to someone (generally to benefit
someone else). But the details of who is harmed, why, for what reason,
and to benefit whom, are all ethical and moral decisions that must be
framed within specific moral systems - concepts your everyday
definition of friendliness fails to capture.

Similarly, would you want a general to have your moral position? (a
general in an "acceptable" war - western allies WWII, for example). A
dominant AI will be much closer to a general or a politician than it
would be to you or me.

Anyway, keep up the experiment, but I feel it has unfortunately little
relevance to AI's as such,


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