[fsml] 'Flesh and Machines': The Future of Robotics (fwd)

From: Eugen Leitl (eugen@leitl.org)
Date: Tue Apr 16 2002 - 06:07:06 MDT

---------- Forwarded message ----------
Date: Mon, 15 Apr 2002 23:35:06 +0100
From: andrew hennessey <a.hennessey@btopenworld.com>
Reply-To: fsml@yahoogroups.com
To: fsml@yahoogroups.com
Subject: [fsml] 'Flesh and Machines': The Future of Robotics

The Nightmare Scifi Future of the Illuminati: Microchips are the first

'Flesh and Machines': The Future of Robotics
April 14, 2002

In 1982 I was sitting in my office trying to convince a British biologist of
the sobriety of the science magazine I edited when my publisher, a woman
wearing a sheer silk scarf in lieu of a blouse, burst in with a large,
flatulent dog on a leash. ''I want you to clone Grundy,'' she said,
indicating the Rhodesian Ridgeback. By way of compliance, I wrote ''Clone
Grundy'' on my desk calendar. She spun and left. The biologist had just
recovered from this interruption when the publisher burst in again. ''Forget
Grundy,'' she said. ''I want you to clone Bob.'' Bob was her husband and the
owner of the magazine. I crossed out ''Grundy,'' wrote in ''Bob.''

Twenty years later, cloning human embryos is almost routine, though it
remains unethical to apply the technology to magazine publishers. So when
Rodney A. Brooks, director of the M.I.T. Artificial Intelligence Laboratory,
tells me that in 20 more years we will have robots with feelings and
consciousness, I'm not going to argue with him.

''Flesh and Machines: How Robots Will Change Us'' is a short, stimulating
book written by one of the major players in the field -- perhaps the major
player -- about the state of robotics and its short-term future. It also
offers surprisingly deep glimpses into what it is to be human. Brooks
appears to have gained a boundless appreciation for human beings by
attempting to copy them.

Traditionally, robots have been built with three loops: a perception
circuit, an action circuit and a ''cognition box'' to provide a worldview
and to mediate between sensors and motors. Brooks challenged this model by
eliminating the cognition box. At a talk, he drew diagrams of bigger boxes
for perception and action -- based on the portion of animal brains devoted
to each -- and made them overlap. He then ''put cognition in a little
cartoon cloud representing the thoughts of an external observer of the
complete system of the robot -- world, perception and action.''

This was radical thought, if not outright heresy. As it turned out, Brooks's
robots didn't need a brain to exhibit intimations of consciousness. The key,
he writes, is to get the robot to react to its sensors quickly. Take
Genghis, Brooks's six-legged robot. It is outfitted with six pyroelectric
sensors, identical to motion sensors that turn on the lights in your
driveway. In Genghis, the sensors are tuned to the infrared band emitted by
the warm bodies of all mammals, and connected to the robot's motors. If a
mammal, say a human, passes in front of Genghis, it moves toward the
infrared radiation. When the mammal stops, Genghis stops. To observers, it
appears Genghis is stalking prey. But stalking implies intent, and Genghis
has none. When Brooks pointed the sensors backward, Genghis walked away from
mammals. Genghis has no internal notion of forward or backward.

Brooks takes issue with the robotics pioneer Marvin Minsky, who has
concentrated on cognition and has belittled the problem of seeing, since
''even stupid people could do it well,'' in Brooks's words. From a robotics
point of view, he suspects that playing chess and solving algebra problems
might be easier to fabricate than visually distinguishing ''between a coffee
cup and a chair'' or walking around obstacles, things that a 4-year-old
child can do. He believes that so-called higher functions derive from the
ability ''to see, walk, navigate and judge.''

Brooks gives us a tour of the marvels of the human vision system and the
difficulties in building them into a robot. One of his brainstorms was
figuring out how to make his robots negotiate a crowded room. He realized
that a robot did not need an ability to avoid obstacles, but could be
programmed ''to seek paths through empty space.'' Zen Buddhist monks spend
several lifetimes learning to grasp the void. Brooks's robots do it as soon
as they are switched on.

His book is full of surprises. Given Genghis's illusory stalking and
Brooks's now famous statement that we ''overanthropomorphize humans,'' I
assumed he was galloping toward the conclusion that consciousness does not
exist. But no. He says consciousness may be ''the result of simple mindless
activities coupled together.'' Our senses and muscles are part of who we

Brooks doesn't paint a bright line between conscious and unconscious, but
implies a continuum of conscious behavior from nuts and bolts to humans,
just as we infer increasing consciousness from lobsters (O.K. to throw in
boiling water) to dogs (non-boilable companions) to chimps (almost human).
When the chess-playing computer Deep Blue beat Garry Kasparov, he said it
played as if it ''had a plan.'' But Deep Blue was qualitatively no different
from chess computers of the 1960's; it was just much faster.

Brooks predicts that we will soon see an explosion of humanoid robots
(first, probably, as house servants) and that we will award them human
rights. He is a persuasive writer, but I worry about his marriage. He cites
his ''darling wife, Janet Sonenberg,'' in the acknowledgments, then says
later that a human is ''a big bag of skin full of biomolecules.'' Good luck,
Ms. Sonenberg.

He is no historian. He finds robotlike machines in antiquity and the
Renaissance, but denies there was any technology in the Middle Ages. In
fact, the Byzantine Empire and the medieval Islamic world were in love with
technology. Tenth-century Baghdad boasted animatronic amusement parks with
singing birds and roaring lions. One of Brooks's themes is that we will
accept robots when we give up the notion of man's specialness, and he trots
out the old saw that the first step came with Copernicus: ''The church could
not accept that the terra firma on which mankind stood could not be the
center of the universe.'' In truth, the church placed the earth at the
center only in the sense that a drain is at the center of a sink: the
detritus gathers there. The intellectual historian Anthony Grafton says hell
was the center of the universe, with the earth circumscribing it. God was
way up above. He was special, not man. You could look it up.

Otherwise, Brooks is thoughtful. He acknowledges the theoretical obstacles
to building robots as smart as their creators. Still, I envision a future
civilization of intelligent robots. They will engage in endless debates over
their origin and the means of their evolution: Natural selection?
Lamarckism? Random drift? A few rusty, gray-bolted robot priests will ignore
all this, chanting two words under their breath: ''Rodney Brooks, Rodney
Brooks, Rodney Brooks. . . .''

* Dick Teresi is writing a history of ancient and medieval non-Western

How Robots Will Change Us.
By Rodney A. Brooks.
Illustrated. 260 pp. New York:
Pantheon Books. $26.


------------------------ Yahoo! Groups Sponsor ---------------------~-->
Buy Stock for $4
and no minimums.
FREE Money 2002.

[FSML] Future Science Mailing List

- To subscribe: http://www.onelist.com/subscribe.cgi/fsml

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:38 MDT