Tracy R Reed wrote:
I knew it! Those lazy programmers. So, Andrew? Why haven't you coded our
strong AI yet? Step to it! Humanity is waiting! :)
It's called Grand Theft Auto IV.
Seriously, the AI actions seem to be really state of the art. People
talk about apologizing to the screen when they accidentally trip
somebody in the game and see the reaction.
But seriously, I've been reading about Lisp and the Lisp Machine and the
Connection Machine and such things lately. All ideas ahead of their time
IMHO. Perhaps their time is coming? The early days of computers sold us
some pretty big ideas and then they turned out to be much harder than we
thought to accomplish and we are still bitter about it. The AI winter is
a very hard pill to take.
No, they were ideas whose time *ended*. They were ideas born of
expensive technology as we moved to cheap technology. Making full use
of scarce resource is important when resource is expensive. Not so true
when things will be 50% cheaper next year.
is: What sort of algorithm might I possibly program that could not be
executed in a traditional serial cpu (although perhaps in greater time)
that isn't just some combination of everything Knuth et al have already
mastered and thoroughly described?
All the heuristic ones. All the ones that are NP. Anything which
gobbles CPU better than O(n^2).
And there are still a *LOT* of those.
A recent article I read talked about how "Twitter" is a fundamentally
hard problem. Everyone has a unique view, so caching doesn't work and
most algorithms for it are O(n^2). Oops.
Part of the issue is that venture money is still trying to make money
off of speculation rather than building new things. Look at all of the
PhD's who went to Wall Street to analyze the stock market movements.
That's a lot of brainpower doing nothing.
It was summed up well by a quote from the same person later in the same
thread:
"Truth is orthogonal to relevance."
-a
--
KPLUG-List@kernel-panic.org
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
