John: Computation is all that there is. But I often try to imagine something
that
is not computation. It depends on different things, and goes into subatomic
physics, string theory, etc.. I think that there are aspects of computation
that we don't understand, and definitely things that I don't understand but
are known among well versed individuals.
Richard: the Ptolemy model of planetary motion imposed a 'simple'
model of the solar system in which everything could be explained by a set
of nested epicycles. There would have been no need to use any other model
because in principle those epicycles could have been augmented to infinite
depth, to yield as good a fit to the data as you wanted.
Has it struck anyone that the Ptolemaic analogy fits AI/AGI rather well?
People hung on to Ptolemaic astronomy because it was so simple - everything
moving in circles - and seemed capable of explaining so much. But it got
desperately complicated to apply and 16th & 17th century folks were acutely
conscious of this. Yet they hung on because of that simplicity.
AI/AGI is also desperately complicated - that's what you keep talking
about - the tangles of "complexity" that it keeps getting snarled up in,
that are an ever present threat, even though they almost never occur in
natural brains.
But, I assume, we all hang on to the system because it is so simple, such
genius - everything, every form of info, is reducible to 1 and 0. The Turing
machine. (Was it Turing who first had the idea of reducing all info to the
same form?) Utter genius. And it works so well for narrow AI.
But everything in life is economic - has costs and risks. And the cost of
computation is there at the very core. Yes it's brilliant to reduce
everything to the same form, but it also obviously creates complications.
Binary is much more complicated than decimal.
Computation as we know it is genius, and indispensible. But it's also -
surely? - rather stupid. If you're always taking longer to work out things
than you really need to, then something is wrong, as well as right. If a
human chessmaster only takes a few hundred steps to calculate a move, and a
computer takes billions to do the equivalent, that, I suggest, should revolt
anyone with any *engineering* perspective. Such waste. An incredibly
intelligent way of being incredibly stupid.
As John suggests, computation as we know it, is unlikely to be the end of
the road. Perhaps we also need something that can handle information
directly, in its native form.
What more do we need? Suggestions? Wild thoughts?
-------------------------------------------
agi
Archives: http://www.listbox.com/member/archive/303/=now
RSS Feed: http://www.listbox.com/member/archive/rss/303/
Modify Your Subscription:
http://www.listbox.com/member/?member_id=8660244&id_secret=103754539-40ed26
Powered by Listbox: http://www.listbox.com