"Robo sapiens" reviewed by Vernor Vinge

From: Michael LaTorra (mike99@lascruces.com)
Date: Wed Sep 20 2000 - 18:28:11 MDT


The review below is from SCIENTIFIC AMERICAN online October 2000 issue at:
http://sciam.com/2000/1000issue/1000reviews1.html

Regards,

Michael LaTorra
mike99@lascruces.com
mlatorra@excite.com

3229 Risner Street
Las Cruces, NM 88011-4823
USA

505.522.5121
............................................................................
............................................................................
..............................................
Apocalyptic Optimism
Menzel and D'Aluisio foresee a superintelligent hybrid species

Robo sapiens:
Evolution of a New Species
by Peter Menzel and Faith D'Aluisio
The MIT Press, Cambridge, Mass., 2000 ($29.95)

REVIEWER_VERNOR VINGE
Computers are the most important thing to come along since ...." It would be
interesting to ask people from over the past 30 years to complete the
preceding sentence. In the 1970s most people might finish the sentence with
"television" or "the automobile" or even "atomic energy." In the 1990s the
ante was raised, and I imagine that the average person would complete the
sentence with "the industrial revolution" or even "the invention of
writing." Now, as the 21st century dawns, some would say: "Computers are the
most important thing to come along since the rise of humankind on Earth."
Even this last comparison may be conservative. Computer hardware power is
doubling every two years or so. If that trend continues, then in another 20
years, our machines will be more powerful than many estimates of human
brainpower.
These supermachines might come about in several ways. In their book Robo
sapiens, Peter Menzel and Faith D'Aluisio examine one of the most
significant and plausible possibilities. The prospect of their title is
explained on the very first page of the introduction: "Robo sapiens: ... A
hybrid species of human and robot with intelligence vastly superior to that
of purely biological mankind; began to emerge in the twenty-first century."
After their excellent introductory comments, the authors indulge in little
speculation. Instead they show us the current state of the art. Photographer
Menzel and journalist D'Aluisio traveled to dozens of research sites around
the world and conducted what must be hundreds of interviews. They have
organized their work into 53 short chapters. Each consists of one or more
interviews, photos of the robots, and sidebars giving project specifications
and background information.
Project by project, they build the overview for us. We see the opinions that
researchers have of the field and of one another's work. We see the conşicts
that exist among different research approaches. No consensus has been
reached as to the path to success or what may be possible. We are in the
midst of an intellectual turmoil, driven by human genius and the steadily
increasing power of our computer hardware.
The breadth of research projects is enormous. We are introduced to
autonomous devices and teloperated ones, uncrewed rovers and robotic
aircraft and pipeline crawlers. Pets and companions, prosthetics and
hands-at-a-distance, housekeepers and welders, weapons and guardians-someone
somewhere has found money (military, commercial, medical) to work on these
possibilities. Many researchers are trying to build machines by imitating
life-and some bioscientists are using robot models to improve their
understanding of life.
It is very difficult to write clearly and simply about such things and still
remain faithful to the underlying technology. Menzel and D'Aluisio
accomplish this feat. The interviews and sidebars are consistently
intriguing but also show great effort in avoiding the misinterpretations
that could come from comparing so many different research enterprises.
(Considering the range of research studied, an appendix categorizing the
different projects would have been helpful.)
Many of the robots are mechanically complex. Often they are unlike anything
we have seen before. Imagine that you had never seen a horse: you would need
clear and objective pictures to get a good idea of the creature's
appearance. Similarly, Menzel's color photos are essential to understanding
the projects described. The photos are beautiful and, for the most part,
extremely effective. In a few cases, however, the higher-level goals of the
photography make the details of the machines unclear-for example, the
motion-blurring effect used in picturing the M.I.T. Spring flamingo robot on
page nine or the Honda P3 robot on page 34 [see opposite page].
The chapters related to robotic pets and artificial affect are especially
fascinating. Setting aside issues of intelligence and mobility, when you pat
your dog on the head and look into its eyes, you can see that someone is
home. Very few of us would say any such thing about a machine. What's
missing from the machine? Perhaps it's simply that our instincts say "warm,
soft, responsive creature with large dark eyes" = "living being with a
soul." Commercial products such as Sony's AIBO and (soon) Hasbro/iRobot's My
Real Baby are early attempts to address this issue. Will people be
attracted, repelled or simply bored by such creations?
The arrival of Robo sapiens is not assured. Maybe we have not yet stumbled
on some basic impossibility. Or maybe our researchers are ignoring some
critical technique: to one degree or another, almost everyone interviewed is
working from the bottom up, imitating the lower levels of animal capability
with a view toward later building the higher functions. The alternative,
top-down approach would seek to model the highest functions of thought
first, perhaps in a supercomputer or computer network-and then, with luck,
the hands and legs and wheels would be easy add-ons. For the past 10 or 15
years, the top-down approach has been in eclipse.
A theme builds through the interviews: living creatures are superior in
almost every way to existing machines. Even some of the superiorities that
we typically grant machines over animals-durability, speed, strength-do not
hold in a natural environment. Similarly, real-time computations that
involve reasoning with the chaos of the material world are also beyond the
computers we have now.
But remember, computer power is doubling about every two years. What we see
now in robotics and AI research is still the historic beginning. Menzel and
D'Aluisio have given us a richly detailed picture of this tumult as it is at
the end of the 20th century. Their book provides a baseline for comparison
with projects in the years ahead. What develops in the next 10 years should
yield strong evidence about where things are going. If we get effective
household robots, if we get vehicles that can navigate real terrain
autonomously, if we get lovable robotic pets ... then we are on track to
create something very like Robo sapiens.

VERNOR VINGE taught mathematics and computer science at San Diego State
University from 1972 to 2000. He is also a science fiction writer; his most
recent novel is A Deepness in the Sky (Tor Books, 1999).



This archive was generated by hypermail 2b29 : Mon Oct 02 2000 - 17:38:37 MDT