[p2p-research] Drone hacking

[J. Andrew Rogers]

On Tue, Dec 22, 2009 at 8:25 AM, Michel Bauwens <michelsub2004 at gmail.com> wrote:
> strange what you say about AI, when I interviewed dozens of AI researhers in
> the late nineties, they said just the opposite, that most of their dreams
> hadn;t come out, that the field had gotten nowhere (but of course they
> expected great things for a hypothetical future, see TechnoCalyps)


By the mid-1990s, the field of AI had run out of ideas. They didn't
call the 1990s the "AI Winter" for nothing. :-)

What saved AI as a field was the development of algorithmic
information theory (AIT) in the mid-1990s. The elephant in the room
for AI over most of its history was that there was no theoretical
basis for it even though it was theoretically plausible, so analyzing
the problem was reduced to trial and error -- an evolutionary
approach. AIT characterized what a very small inductive machine could
predict with respect to a very large data environment (e.g. Feder et
al 1992) which was eventually noticed by a few people working on AI as
astonishingly relevant to their work.

AIT has proven to be an excellent basis for describing the fundamental
problem in a way that was not possible before. Importantly, it showed
in rigorous terms why the history of AI was one long string of
failures *and* that a real solution would look nothing like what had
already been tried.

-- 
J. Andrew Rogers
realityminer.blogspot.com