From: Nick Tarleton (nickptar@gmail.com)
Date: Wed Mar 12 2008 - 17:13:29 MDT
On Wed, Mar 12, 2008 at 6:38 PM, <mwaser@cox.net> wrote:
> > Er, what exactly *is* this formulation? Is it explained in a paper
> > elsewhere? You didn't provide a link.
>
> Sorry. The formulation *is* the declaration that was attached with the most important part being that the primary overriding goal of a Friendly entity is . . . . "Do not act contrary to someone's/anyone's goals unless absolutely necessary for the fulfillment of a reasonable/rational personal goal (explicitly not including generic sub-goals like money, power, pleasure, religion, etc.)"
So what *does* it include?
> Pretty much everything derives from there and the rest of the declaration is mostly administrative details handling protection against UnFriendlies falsely declaring themselves to be Friendly.
>
> If you believe that I can stay true to the declaration AND still do something that is both horrible and unethical, please let me know (because that means that my theory has a hole in it).
Without a definition of "reasonable/rational personal goal", how can
anyone say? Even then, AFAIK you haven't explained why it's in every
AGI's self-interest to be Friendly.
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT