[ExI] Universally versus 'locally' Friendly AGI
stefano.vaj at gmail.com
Tue Mar 8 23:19:51 UTC 2011
On 8 March 2011 23:13, Samantha Atkins <sjatkins at mac.com> wrote:
> This must be tongue in cheek. A true AGI may very well flunk the Turing
> Test. Not by being too stupid or being less capable but by being unwilling
> to dumb itself down to such an asinine level. I doubt very much that
> passing as human will be high on the priority list.
If passing the Turing test and emulating bio ethology is not a
requirement, I wonder how we can distinguish an AGI from any other
device (or natural phenomenon, for that matter) exhibiting universal
computation features, as defined in A New Kind of Science.
> Stefano Vaj wrote:
>> As a consequence, AGIs will probably not be the most intelligent
>> entities around. They will simply be those amongst them who take
>> themselves as humans or as humans' children.
> Then why even bother. If AGIs are not much much better than that in
> potential then whatever is the point?
Because it is an interesting experiment per se, or because we might
identify with persuasive enough emulations of our identity and be
willing to leave them behind and/or around for instance?
But I agree that some overhype exists on the very concept. A computer
may be much more powerful than an AGI without exhibiting typical AGI
features, and there is nothing in principle which can be done by an
AGI that cannot be achieved by some other program, and/or with the
assistance of a biological brain.
More information about the extropy-chat