MT:There isn't an
AGI system that has
shown, in even the most modest way, higher
adaptivity - the capacity, in any
given activity,  to find new kinds of paths, or take
new kinds of steps, to
its goals - which are, by definition, not derived
from its original
programming. The capacity, say, to find a new kind
of path through a maze or
forest.

Tom McCabe: Pathfinding programs, to my knowledge, are actually
quite advanced (due primarily to commerical
investment). Open up a copy of Warcraft III or any
other modern computer game and click to make a
character go from one end of the map to the other. How
does it find a correct route? Pathfinding AIs.

I said an AGI must have the capacity to find a "new kind" of path - as animals have done throughout evolution. Just finding your way from one end of a map to another doesn't qualify. We don't call people "pathfinders" if that's all they do.

But this failure to distinguish between basic adaptivity and higher adaptivity - or, if you like, iterative and creative pathfinding - - runs right through AGI to my mind..

Tom:By the time the AGI has enough intelligence to say
"Hi!", I'm betting that at least 50% of the work will
already be done.

Er, when your AGI says "Hi" to someone, somehow I don't think the world is going to say "Hallelujah, AGI has arrived." If you and others can even think for a second that's 50% of the work - no wonder people are so extremely casual in estimating AGI's arrival.

P.S. For an example of simple but creative pathfinding, take UK birds recently who suddenly decided to switch from their normal dead reckoning flight path for long journeys to following the road highways instead. You've got to be able to go off the map to put AGI on the map. {That last sentence is also a form of higher adaptivity - if your AGI could say something like that rather than "Hi" - it would certainly have arrived).




-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=4007604&id_secret=27068450-9d3434

Reply via email to