Re: and visual library once again

["Bob Spooner" <rls19@psu.edu>]

"Steve Whalen" <SteveWhalen001@hotmail.com> wrote in message
news:1130224998.944468.133100@g44g2000cwa.googlegroups.com...
> Bob Spooner wrote:
> > "Steve Whalen" <SteveWhalen001@hotmail.com> wrote in message
> > news:1130049078.633311.55000@g44g2000cwa.googlegroups.com...
> > >
> > >  <snip>  Much the way the DARPA
> > > challenge of having an autonomous vehicle drive itself over 100 miles
> > > through rough desert terrain was solved: not by elegant software, but
> > > by practical engineering and brute force computing. In the time-frames
> > > I was referring to (50 to 100 years from now), the power of today's
> > > most powerful weather simulation super computers will fit inside a
> > > hollowed out grain of sand.
> > >
> > I think you would be surprised at the elegance of some of the artificial
> > intelligence techniques that are used. Where I work, we do research on
and
> > with autonomous vehicles. The brute force approach simply doesn't work.
> >
> > Bob
>
> I'm sorry. I didn't mean to denigrate or downplay the work being done
> in the field (or the work that was done for the DARPA challenge). I'm
> not at all surprised that there is some very elegant work in AI being
> done for autonomous vehicles.
>
> Doing the kind of hard work you do was considered unnecessary and
> "brute force" by many AI pundits of the 1960's who were going to
> (within 10 years!) emulate the human brain in a computer and then just
> have the computer brain learn do any task a human could do.  Marvin
> Minsky's book I referred to was basically saying that was b.s. and
> wasn't going to happen (which I agree with, then and now).
>
> About every 10 years or so someone from the AI community says something
> similar...  That's the crowd I was thinking of, to whom anything other
> than putting a human brain in a computer and letting it figure out how
> do something, is "brute force"(I'm oversimplifying again: it was quite
> a battle for $$ and research grants and reputation and ego: thankfully
> that never happens any more <g>).
>
> We may get closer to being able to emulate a human brain in a hardware
> computer in 100 years, but I suspect that most if not all of the
> "useful" AI work that will power the robots and such I was talking
> about, will come from the kind of work you do. I still think we should
> continue to learn from the brain, but I think trying to program a
> hardware computer to emulate how a human brain works isn't going to
> magically make a "smart" computer.
>
> I just hope Ada (or it's successor) is still around in an hundred years
> and being used for the more critical components of the robots we
> entrust our life to. Ironically the last time I did any serious
> robotics about 10+ years ago, Forth was my language of choice for the
> task at hand due to memory and hardware constraints. I guess that's a
> part of the point I was trying to make: "hardware constraints" are
> beginning to disappear, and will have disappeared completely in 100
> years.
>
> Steve
>

Well, I don't think that hardware constraints will ever completely
disappear. That's because of what I call "Spooner's law of computing" which
says that computing power generates the need for more computing power.

Bob