Re: More silly but friendly ideas

From: Vladimir Nesov (robotact@gmail.com)
Date: Tue Jun 03 2008 - 14:55:00 MDT


On Wed, Jun 4, 2008 at 12:11 AM, John K Clark <johnkclark@fastmail.fm> wrote:
> "Vladimir Nesov" robotact@gmail.com
>
>> your thesis is that it just can not be done, period.
>
> It can't be done and shouldn't be done. It is imposable, evil, and above
> all silly.
>
>> if I come up with a simple theory of Friendliness,
>> I can as well act in the capacity of Oracle AI
>> in this scenario.
>
> I you can come up with a simple theory of Slavery then why do you need
> the AI in the box?
>

Then I won't need it. Creating an Oracle AI is strictly easier than
creating Friendly AI (although arguably not by much), that is why the
concept looks useful. Oracle AI only needs to be cooperative in
short-term and in retrospect, while Friedly AI needs to stay so
indefinitely. I can't guarantee that I will be Friendly indefinitely,
but I'll do just fine as an Oracle AI if I somehow manage to produce a
theory of Friendliness.

-- 
Vladimir Nesov
robotact@gmail.com


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:03 MDT