Re: Why would AI want to be friendly? (Was: Congratulations to Eli, Brian ...)

From: xgl (xli03@emory.edu)
Date: Fri Sep 08 2000 - 08:42:13 MDT


On Wed, 6 Sep 2000, phil osborn wrote:

> 1. All intelligence is fallible.
> 2. Conscious intelligence uses perceptual feedback for base level
> correction.
> 3. Emotional interaction with other conscious beings is the only method by
> which one can "mirror" one's consciousness, as such.

        i'm not sure what you mean by "mirror" ... why is it necessary and
why must emotional interaction be the only means to achieve it?

> 4. Acting as a predator makes real emotional interaction impossible, as one
> has to conceal part of ones nature and this required putting a conceptual
> process into the loop, defeating the epistemological purpose - perceptually
> reaffirming the nature of one's consciousness. (This is one of the base
> reasons polygraphs work - and why we place such an extravagent value on
> things like love.)

        i always thought polygraphs work because human brains are running
buggy ware. as to the "epistomological purpose," what confirmation have i
that it is as fundamental as you claim?

> 5. Any smart conscious being is capable of figuring this out.
> Therefore, if the SI is conscious, we probably don't have to worry -
> although I would want to have a serious talk with it on this very subject
> before I put it on line. A very smart, non-conscious SI is another matter.
>

        i'm not very smart, so i have yet to figure this out; and i
suspect there are many like me.

-x



This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:30:52 MST