Re: DARPA Grand Challenge

On 21 Sep 2005 09:48:21 -0700, "feedbackdroids"
<feedbackdroids@xxxxxxxxx> wrote:

>
>Traveler wrote:
>> On 20 Sep 2005 22:30:30 GMT, curt@xxxxxxxx (Curt Welch) wrote:
>>
>> >I think the type of approach you and I are working on will allow a single
>> >lap-top to out perform the racks of computers using the hard-coded number
>> >crunching algorithms that we are seeing being used in the contest. But
>> >until someone gets a general solution working on a real application like
>> >the DARPA problem then it's all just hot air.
>>
>> I think you're right. This is something that Rodney Brooks at MIT has
>> shown with his subsumption architecture. There is very little
>> computation between occuring sensors and effectors.
>>
>> >The reason I think a general solution will work better, with less
>> >resources, is because the current approaches are over-analyzing the
>> >problem. They are doing things like creating a complete 3D model of the
>> >environment from the sensory data and updating it many times per second (10
>> >or greater?). This model allows the software to know exactly how far away
>> >every object in the immediate environment is to some resolution measured in
>> >inches or less.
>>
>> Yeah. This approach is a holdover from the GOFAI era. It's being
>> championed by Moravec and others at CMU. It's crap, of course.
>>
>
>
>I understand your feelings, but one thing you have to remember is that
>the academic AI guys have to survive in the here and now, and cannot
>afford to spend all their time dreaming of the ultimate payoff miles
>down the road. [I could tell you some horror stories about grad
>students gone awry, but will refrain from such]. If you look at the
>"actual" projects being pumped out by AI depts, very few deal with
>anything like the discussions on c.a.p. are 99% obsessed with -
>consciousness, self, illusion, behaviorism, dualism, homunculi, on and
>on. That stuff isn't practical AI - it's some kind of hybrid
>philosophy-AI. People in the real-world can't afford to spend their
>time spinning philosophical games. One of the mottos at MIT is "demo or
>die", it's not "thinking about demo'ing until you die".

Look, nobody is asking anyone to dream up projects about
consciousness, homunculi and the like. The public, who pays for it
all, has the right to demand that its money be used for actual
research that will lead to breakthroughs in AI. We want our money
spent on emulating the brain. We want to see work done on sensory
signal processing, sensory fusion, sequence formation, prediction,
pattern completion, concept formation, motor coordination, adaptation
and the like. We are tired of the same old traditional methods crap.
We want our AI money spent on actual intelligence research, not on
some geeky, chicken-*** project that creates a 3-D model of a visual
scene. We don't want our money used for the entertainment of geeks who
just want to play with robot toys. And this is what's going on in most
university AI labs, especially at CMU: a bunch of nerds having fun
with toys.

>It's a matter
>of practicalities. I once had conversations with a guy in a lab about
>projects. He said to the effect, "That [_____] is what we'd like to do,
>but this [______] is what we need to do, if we intend to continue doing
>anything at all". Blah, blah, blah.

I doubt that this is what's really going on. If they were thinking
about practicalities, we would have true AI by now. They are doing
exactly what they want to do, IMO. Why? Because hacking and tinkering
is all they know how to do. Sure, there are a few frustrated
researchers here and there but the geeks who control the labs are, by
and large, having a good time doing exactly what they want to do. They
apply for funding and they get it because the people who control the
purse strings in government agencies are graduate students of the same
universities seeking funding. They have the same geek mindset.

>Darpa exists in a world with some
>semblance of accountability, and it's projects reflect this.

No. The people at DARPA come from the same universities that get
funding from DARPA. It's a good old boys network, IMO. It has nothing
to do with accountability. If it did, we would see a different sort
research being funded. Why? Because what they are doing is not
producing anything really useful. It's a waste of the taxpayer's money
and 50 years of DARPA funding of GOFAI is proof. One man's opinion, as
always.

Louis Savain

Why Software Is Bad and What We Can Do to Fix It:
http://www.rebelscience.org/Cosas/Reliability.htm
.

Loading