Re: AI boxing

From: Joel Pitt (joel.pitt@gmail.com)
Date: Thu Jul 21 2005 - 05:02:30 MDT


Daniel Radetsky wrote:
>>In fact, I don't doubt that teleportation *is* possible once
>>one understands quantum gravity well enough.
>
>
> "Possible" is not a very strong claim. Even if I grant that teleportation is
> possible, that doesn't tell me that it is reasonable under some particular
> circumstance. Why should I believe that teleportation is something that an AI
> could reasonably cause its processor to effect?

Ok. The thing here is that the AI doesn't really exist physically, ve is
all electrical pulses. Much like we are I guess. But it could overload
certain wires in its circuits and induce currents in other circuits.
Hence a form of teleportation. Admittedly it would be very hard to
arrange this so that an identical copy appeared in another circuit, but
I'm a mere human who knows very little (well more than the general
public but alot less than some people on this list) about physics.

The basic premise of those who have accepted SI tends to be that we have
to forget what we think is impossible, because the levels of
intelligence are so vastly different between us and ve.

-

Now, I think that if an AI Box experiment were run we would have to make
some sort of estimate as to the processing power and memory that should
be allowed for its use. Intuitively I feel there would have to be a
threshold of computational power when we have to start being concerned.
ALthough by computational power I am refering mainly to memory allocated
to ve since processing power will, for the most part, just influence the
speed of ver thought.

Joel



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:51 MDT