From: Stuart Armstrong (dragondreaming@googlemail.com)
Date: Sat Apr 19 2008 - 14:40:29 MDT
> > Schematically:
> >
> > 1) The decision is made by the programmers.
> > 2) The decision is made by a small group of people with their own
> > interest in mind.
> > 3) The decision is made by a small group of altruistic people.
> > 4) The decision is made by some sort of democratic process, strongly
> > guided by those with a good understanding of the issues.
> > 5) The decision is made by some sort of pure democratic process.
>
>
> Extrapolated volition? (In which the initial dynamic comes from (3),
> and the end result from something like (4) or (5) but without at least
> some of democracy's flaws.)
I'm sceptical about the whole extrapolated volition concept. My
objections are properly written up in
http://www.neweuropeancentury.org/GodAI.pdf
But to paraphrase: even those advocating extrapolated volition
acknowledge that there could be problems (and sometimes allow a "final
judge" to check up specific things, like whether the extinction of
humanity is on the cards). But if EV can go "wrong" in one way, it
could go wrong in many ways - ways we may not even conceive of now,
but if we were told about, we would say "that's nearly as bad as going
extinct!". So EV has to be throughly checked before being implemented
as a moral basis; a final judge with one or two specific isn't enough.
So we haven't gained anything. Decisions still have to be made. How?
Stuart, who would not trust his own EV without strong evidence
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT