Re: [NTLK] Newton History

On 21.02.2009, at 10:30, Larry Yaeger wrote:
> Pretty much everything else is doable, or done, and I don't doubt
> that human-level AI will be with us someday, but A) it's a long ways
> away, and B) I can't help but wonder if anything will ever be both
> that intelligent and that pliable--genuinely smart, yet
> unquestionably devoted to giving you a better computer interface its
> entire artificial life. It's a strange, science-fictional thing to
> contemplate, but if a machine is as smart as a human, forcing it to
> do menial tasks is nothing more than slavery. But, then, who knows,
> maybe it will be possible to tailor intelligence into forms that
> aren't self-aware and don't demand their individual freedoms--I kind
> of doubt it, but we really don't know. However, if they are capable
> of and do demand their rights, it would be immoral to deny them.

Just to give a little impression from the other side of "servants": I
have started building robots a while ago, and even devoted a year a
full time work to create a "walking robot" kit that could be assembled
at home. The idea was to create a starter kit, and then produce more
elobarate mechanics and taller bipeds with each iteration, eventually
reachin 4 feet, so those robots would actually be able to help in a
household.

A baby was on the way, and my goal was to keep my robots growing at
the same pace as my daughter was growing. I though that 12 years to
create a functional 4ft tall robot should be enough. After all, a
biped is just a collection of joints and motors, and with access to
laser cutting, CNC, modern plastics and carbon tubes this should be a
breeze, right?

Boy was I wrong. The human body is an insanely complex system evolved
over millions of years into what it is now. All our integrated
circuits, electric motors, and high strength alloys can barely compose
a similar "device". Sure, I can make a 40 cm electric bipedal
contraption, and program it to stand up and walk, but even Honda and
their Assimo after many years and great spendings is still a klutz,
boxy, uneven, and basically dumb.

But why can a five year old run, fall, jump, be goofy, fall again and
not get hurt? Whenever the 40 cm robot falls after a few clumsy steps,
a gear will break, a joint will bend, and the calibration will be off.
What is it that makes our dream of the helper robot remain a dream for
hundreds of years now?

It's the mind that makes our old bone sacks work! Forget about the
mechanics. The mind takes in an unbelievable mass of information,
computes it, and outputs another huge amount of commands that teach
our body to stay out of trouble.

As long as we are not able to build a machine that runs arond the back
yard, jumps off a swing set, climbs a rope ladder, and knows how to
master a seesaw and build sand cakes, we are still decades away from a
computer assistant at PhD level ;-)

Matthias

----
http://robowerk.com/
==================================================================== 
The NewtonTalk Mailing List - http://www.newtontalk.net/
The Official Newton FAQ     - http://www.splorp.com/newton/faq/
The Newton Glossary         - http://www.splorp.com/newton/glossary/
WikiWikiNewt                - http://tools.unna.org/wikiwikinewt/
====================================================================