There were three recent books on computing that got reviewed together=20
in the NYT & elsewhere:
* When Things Start to Think; Neil A. Gershenfeld
* Robot : Mere Machine to Transcendent Mind; Hans P. Moravec
* Age of Spiritual Machines : When Computers Exceed Human=20
Intelligence; Ray Karzweil, Ray Kurzweil
Kurzweil's book:
http://www.amazon.com/exec/obidos/ASIN/0670882178/o/qid=3D917053114/sr=3D2=
=20
-1/002-1428174-4307411/forkrecommendedrA
first chapter=20
http://search.nytimes.com/search/daily/bin/fastweb?getdoc+site+site+57=20
815+4+wAAA+Kurzweil
Slate did a joint review of these books, attached. at the very end is=20
a mention of
"His image of a "utility fog"--a set of microscopic machines, acting=20
in unison to produce objects atom by atom on demand--is really a=20
version of the "holodeck" on Star Trek. The utility fog goes one step=20
beyond modeling reality: It makes it. But the link to that fantasy=20
comes from computer games."
The NYTimes Review quoth:
"If our three authors are wobbly on the philosophy of mind and=20
artificial intelligence, they are strong on computer technology=20
itself; and here is where their books are particularly interesting.=20
The reader can simply detach all the dubious speculations about=20
machine consciousness and focus on the authors' predictions about the=20
future of computer and robot technology, its potential benefits and=20
hazards. Consider two examples of the kind of technology that might=20
well be just over the horizon: the foglets and the nanobots. Foglets=20
are tiny, cell-sized robots, each more computationally powerful than=20
the human brain, that are equipped with minute gripping arms that=20
enable them to join together into diverse physical structures. At=20
ease the foglets are just a loose swarm of suspended particles in the=20
air, but when you press a button they execute a program for forming=20
themselves into an object of your choosing. We may come to live in=20
foglet houses whose rooms are formed from the same foggy swarm. We=20
may come to have foglet friends and take foglet vacations. Our entire=20
physical environment may come to consist of a 3-D mosaic of=20
cooperating microscopic computers. This would be virtual reality made=20
concrete. "
http://www.nytimes.com/books/99/01/03/reviews/990103.03mcginnt.html
Here's a monograph on the concept:=20
http://nanotech.rutgers.edu/nanotech/Ufog.html
Sussman's amorphous computing talk : http://world.std.com/~wware/sussman.htm=
l
a/k/a foglets in Drexler's Nanosystems and Smart Dust in PARCisms.
The critical inspirational text: Drexler's _Engines of=20
Creation/Engines of Destruction_. This is a nice hook for the=20
social-responsibility aspect of Engineering: the role of the field to=20
engender innovation, as well as governance and ethics. Nanotechnology=20
can clearly be more destructive than atomic weaponry, and Drexler=20
doesn't shy away from it:
http://www.amazon.com/exec/obidos/ASIN/0385199732/o/qid=3D917054230/sr=3D2=
=20
-1/002-1428174-4307411/forkrecommendedrA
More to come...
Rohit
----------------------------------------------------------------------=20
----------------------
David Bennahum is the author of Extra Life: Coming of Age in=20
Cyberspace. Jim Holt writes about science and philosophy for Lingua=20
=46ranca and the Wall Street Journal. This week they discuss When=20
Things Start To Think, by Neil Gershenfeld (Henry Holt; 224 pages;=20
$25) and The Age of Spiritual Machines: When Computers Exceed Human=20
Intelligence, by Ray Kurzweil (Viking; 352 pages; $24.95).
Previous List all Next
=20
1/4
From: Jim Holt
To: David Bennahum
Subject: An Earnest Question
Posted: Monday, 10:32:00 AM PDT
Dear David,
Let me begin with an earnest question: What good are computers? A=20
reasonable wish is that they might help ease Adam's curse by, say,=20
enabling us to produce more goods and services with less labor. So=20
far, though, they have failed to do this. The enormous investment=20
that American businesses have made in computers over the last couple=20
of decades has yet to yield any detectable economic dividends. As one=20
economist put it, "You can see the computer age everywhere but in the=20
productivity statistics." The best estimate of the contribution=20
computers have made to economic growth is a miserable .15 percent a=20
year. (See, for instance, Daniel E. Sichel's The Computer Revolution:=20
An Economic Perspective, the Brookings Institute Press, 1998.) This=20
is quite embarrassing when set beside the hyperbolic claims of Al=20
Gore, Alvin Tofller, and George Gilder, who insist that the computer=20
revolution is as momentous as the earlier transition from an=20
agricultural to an industrial economy.
Computers have made possible the "information highway," but what good=20
is that? Very little information--some literature, some science--is=20
intrinsically valuable. The rest is valuable only instrumentally and=20
differentially, insofar as I have it and you do not. When such=20
information becomes universally available its value drops to zero.
Since computers have so far been a disappointment when it comes to=20
making better and cheaper widgets, it is not surprising that those=20
who are existentially committed to the machines should retreat to=20
spiritual grounds. Thus the appearance of books like the two we are=20
discussing, which argue that lying just over the horizon is a golden=20
age in which computers will become our "mind children." Our=20
mechanical creations will presently exceed us in intelligence and=20
blossom into full self-consciousness, we are told. By an ill-defined=20
process called "downloading" (or sometimes "uploading") we will be=20
able to extract our personal essences from the piece of meat in our=20
skulls and install them in the new generation of computers, thereby=20
obtaining immortality, not to mention better sex. All of this will=20
happen in the next 50 years--that is guaranteed by the "exponential=20
law of computing."
Such extravagant claims for the metaphysical potential of computers=20
have a slightly moldy odor about them. Way back in the Eisenhower=20
administration, the Nobel laureate Herbert Simon, one of the fathers=20
of Artificial Intelligence, was already claiming that by writing=20
computer programs that could deduce theorems in elementary geometry=20
he had resolved the mind-body problem. Even before Watergate and=20
disco, Marvin Minsky, another patron saint of AI, proclaimed that "In=20
from three to eight years we will have a machine with the general=20
intelligence of an average human being." Okay, Deep Blue did beat=20
Kasparov in chess, but no computer today, however festooned with=20
sensors and servo-mechanisms, can do what any child can: move the=20
chess pieces from square to square. Nor can the most sophisticated=20
software do so much as understand a simple story in English.
Given AI's history of hype and unrealized sci-fi fantasy, we can, I=20
think, be forgiven for taking such futuristic visions in these two=20
books with a pinch of salt. Or can you, David, give me a reason for=20
thinking that computers really will add "extra life"? I ask with no=20
ulterior prejudice, but in the sprit of pure inquiry.
=20
1/4
From: David Bennahum
To: Jim Holt
Subject: The Ultimate Sandbox
Posted: Monday, 3:21:00 PM PDT
Jim:
Thanks for your note about Ray Kurzweil and Neil Gershenfeld's books.=20
While I agree with many of your criticisms of these two books, I'm=20
amazed that they triggered this "Computers! What are they good for?=20
=2E.. Absolutely nothing!" polemic on your part. I can't even begin to=20
discuss the nuttiness in these books without first dealing with the=20
nuttiness of your anti-computer sentiments.
Computers have substantially improved our lot. Without them, the=20
complex systems we take for granted would not exist: medical expert=20
databases doctors use to diagnose illnesses and treatments,=20
air-traffic control, the entire telephone network with its collapsing=20
cost of communications, jet-travel, weather forecasting that saves=20
lives, pharmaceutical development using computerized models, plus=20
intensive historical studies of ecological damage, worker safety,=20
education patterns, and so on that are used for private and public=20
policy making. All of these depend on computers for a simple reason:=20
Computers are modeling machines. They build models of reality we can=20
then use for testing, saving us from the dire choice of only learning=20
through real-world experience. The computerized model of America's=20
skies is what allows jet planes to fly without hitting each other. In=20
short, without computers we cannot have complex systems, and without=20
complex systems, we cannot have as many choices in how we travel and=20
communicate. That's just one area where computers have shown some=20
good.
Your criticism is that computers haven't done much good because=20
worker productivity hasn't gone up. But who says that worker=20
productivity is the measure of a computer's usefulness as a tool?=20
It's just one measure, and one I don't particularly like, since I=20
believe we all have the right to laze about on the job every now and=20
then, and if computers make it easier to have fun while getting paid,=20
I'm all for it. Which gets me to my second point: Computers are good=20
for entertainment. Whoever is reading this message right now--I hope=20
you're having a good time reading it, which proves my point. One of=20
my favorite computer concepts is the "Here's the Boss!" button.=20
That's a special extension which, when the right combination of keys=20
are pressed, instantly produces a spreadsheet, covering up whatever=20
else you might be doing at that instant. The reason we need "Here's=20
the Boss!" buttons is because computers--especially computer games,=20
personal electronic mail, and personal web sites--are good at=20
modeling things that give us pleasure, be it by recreating the=20
written word in electronic format, or the labyrinthine levels of DOOM.
You wonder if the "Information Highway" serves any purpose, since=20
"Very little information--some literature, some science--is=20
intrinsically valuable. The rest is valuable only instrumentally and=20
differentially, insofar as I have it and you do not. When such=20
information becomes universally available its value drops to zero."=20
To this I can only say: GET AN E-MAIL ADDRESS, JIM! I must tear the=20
veil, and reveal that while I write this in my Eudora mail=20
application, Jim is in the woods of New England dealing with a=20
typewriter, fax machine, and maybe a printer of some kind. Do you=20
have outdoor plumbing yet? I say this because the value of the=20
information highway has little to do with information, and more to do=20
with communication. It is a communications medium, and what makes it=20
so marvelous is that it is the only instance in all of human history=20
of a medium where it is as easy to reach one person as it is to reach=20
thousands or millions. I see it as an antidote to our broadcast=20
media, where the ability to send to many is strictly controlled. It=20
is an antidote to a consumer culture where we are expected to consume=20
information, but not produce it (except as data bits in a credit=20
report). E-mail and personal home pages (80,000,000 Americans now=20
have e-mail access) are why TV watching is down for the first time=20
among teenagers in the entire history of TV. For that alone,=20
computers have surely benefited humanity.
But none of this excuses the turgid texts we both wrestled with over=20
the weekend. Gershenfeld and Kurzweil offer variations on geek=20
Valhalla. For Gershenfeld, a director at the MIT Media Lab, and head=20
of a project called Things That Think, what's wrong with computers is=20
twofold: Their interface is mediocre (all those windows, icons, and=20
mice), and they need to be connected to everyday things, like shoes,=20
to make them more useful. For Gershenfeld, the computer exists as a=20
helot, albeit a helot without any limbs or ability to interact with=20
its environment. When Things Start to Think is an ode to getting=20
computers hooked up to physical objects, giving "things" the ability=20
to "think." It's a very Media Lab-ish precept, one both mundane in=20
its ambition--it's about making better digital butlers--and familiar=20
in its anthropomorphism. For Gershenfeld, computers are in a fetal=20
stage. His job is to turn them into children, able to teach and learn=20
for themselves.
Kurzweil blows past the child stage of development, and goes straight=20
for quasi-Buddhist vision of using computers to break the cycle of=20
birth and death, using them to reach a kind of immortality. In=20
Kurzweil's Spiritual Machines he makes a familiar argument: Bodies=20
are bad, and our human "wetware" must be shed, while our=20
brains--"software"--float ascendant, finally freed from the prison of=20
flesh and mortality. Where once transcendence came from meditating=20
and not eating for a long time, Kurzweil posits that computers,=20
following an inevitable track of Darwinian self-perfection and=20
selection (what about mutation?), will soon become the symbiotic=20
vessels for our mutual ascendance to semi-omniscience--godhood.
Kurzweil is really the latest iteration of earlier books like Bruce=20
Mazlish's The Fourth Discontinuity: The Co-evolution of Humans and=20
Machines (Yale; 1993) and Gregory Stock's Metaman: The Merging of=20
Humans and Machines into a Global Superorganism (Simon & Schuster;=20
1993). These earlier authors, much as Kurzweil does, argue that=20
computers and humans are merging into a new organism. Kurzweil makes=20
the same argument, but more squarely places it in the context of=20
"spirituality." But like his predecessors, Kurzweil skirts painfully=20
close to the creepily genocidal, or at least all-obliterating. When=20
surveying the glorious future, Kurzweil throws out one liners that=20
reflect the cavalier, computer-game-like (SimFuture) esthetic behind=20
this sort of writing: "If we're going to enter a new world, we had=20
better get rid of traces of the old" (p. 142). Kurzweil throws that=20
out in the middle of a rapturous discussion of future "virtual=20
reality." What is "get rid of traces of the old" supposed to mean=20
anyway? For both Kurzweil and Gershenfeld, history does not exist.=20
They write of portentous human change with little reference to the=20
past (icky, "old" stuff), other than the typical rehashes of=20
chess-playing automata housing dwarves from the 18th century, Charles=20
Babbage, Turing, and Marvin Minsky. Talking about history mucks up=20
the fantasy, so it's best left alone.
Kurzweil's book is far more ambitious than Gershenfeld's, which in a=20
way is a compliment to Gershenfeld, who seems to treat humanity with=20
a measure of dignity, and appropriate non-messianism. For=20
Gershenfeld, being human is okay, and computers are just here to help=20
out. Kurzweil, in his argument that computers are following an=20
inexorable slope of ever increasing computational power, which must,=20
around 2020, lead to a tipping point where they become conscious=20
(because they'll have the same computation ability of the human=20
brain), is essentially arguing through a combination of metaphor=20
(brains are like computers) and leaps of faith (when the density of=20
connections in a chip is as dense as a brain, something clicks into=20
smartness). But of course none of this has been proven. It's science=20
by analogy, and only Albert Einstein knew how to do that well. But=20
then again he was a genius.
At the end of your message, Jim, you asked me: "can you, David, give=20
me a reason for thinking that computers really will add "extra life"?=20
I ask with no ulterior prejudice, but in the sprit of pure inquiry."=20
As machines for fantasy, play, and communications, computers=20
certainly have given me extra life--the phrase, btw, comes from=20
computer games, where high scores earn "extra life" for extended=20
play--in the sense of my having discovered a wonderfully creative=20
co-conspirator. Because computers are just modeling machines, they=20
exist as the ultimate sandboxes for human fantasy. What other device=20
can be a movie, book, radio, telephone, mailbox, newspaper, and=20
immersive game? Whether they're any good at building cheaper widgets=20
is General Motors' problem, not mine. I know they're good at making=20
me happy.
/d
David Bennahum is the author of Extra Life: Coming of Age in=20
Cyberspace. Jim Holt writes about science and philosophy for Lingua=20
=46ranca and the Wall Street Journal. This week they discuss When=20
Things Start To Think, by Neil Gershenfeld (Henry Holt; 224 pages;=20
$25) and The Age of Spiritual Machines: When Computers Exceed Human=20
Intelligence, by Ray Kurzweil (Viking; 352 pages; $24.95).
Previous List all Next
=20
1/5
From: Jim Holt
To: David Bennahum
Subject: Our Duty to Mock
Posted: Tuesday, 11:17:00 AM PDT
Dear David,
I began by raising the possibility that computers have not=20
contributed to the sum of human happiness. There are two components=20
to this sum: an economic one that can be measured in dollars, and a=20
non-economic one that we can only make plausible conjectures about. I=20
produced some hard evidence that the computerization of the American=20
economy that has taken place over the last decade has not done=20
anything to boost the GDP (as witnessed by the fact that productivity=20
growth in the '90s hasn't even reached the anemic rate of the 1950s,=20
much less that of the '60s and '70s). You counter with anecdotal=20
evidence that computers have yielded some economic benefits.=20
Apparently these are almost completely neutralized by computer-caused=20
wastage: rapid obsolescence, excessive training costs, otiose e-mail,=20
and a general profusion of not very valuable information. The=20
hypothesis that computers have substantially bettered our economic=20
lot has been tested empirically and found to be false. Or do you=20
dispute the econometric methodology?
On to the non-economic component of the sum of human happiness.=20
"Computers are good for entertainment," you say, adducing=20
(improbably) this book discussion as a case in point. I cannot argue=20
with your subjective experience, but surely computers must occupy a=20
pretty lowly spot in the entertainment hierarchy, well below playing=20
the piano, skiing, drinking claret with friends, reading a novel, or=20
enjoying a one-night stand. Nor have computers done anything to=20
improve culture at large; neither music nor art nor literature has=20
got better; cinema and political journalism--thanks to special=20
effects and the Internet, respectively--have been demonstrably=20
degraded. And there are few things more dispiriting than the sight of=20
spotty and stupefied teenagers sitting for hours in front of computer=20
screens at the local library while great volumes of Web detritus=20
scroll by. As a communications medium, computers have allowed=20
individual nutcases, once kept isolated from one another by accidents=20
of geography, to come together and for virtual nutcase societies that=20
can do actual harm.
The unintended message of the two books we are discussing--especially=20
Kurzweil's--is that it's only going to get worse. I did not find=20
Kurzweil's book "turgid," as you did. As a would-be philosopher of=20
consciousness he has little to say that is original, and sometimes he=20
seems to make thinks up ("The late Wittgenstein heavily influenced=20
the existentialists"--who told him that?). He betrays his tin ear for=20
culture when he observes, for example, that the harpsichord is an=20
"obsolete technology."
But when Kurzweil is conjuring up visions of the next century, his=20
book can be fascinating in a ghastly sort of way. He loves computers=20
so much, finds them so entertaining as a co-conspirator for fantasy,=20
that he is positively lusting to merge with them. Thanks to=20
nanotechnology and the "partial porting" of our brains with computer=20
devices, he predicts, the world will become utterly transparent to=20
our desires in the sweet by-and-by. He writes,
It may seem that we will have too many choices. Today we have only to=20
choose our clothes, our makeup, and destination when we go out. In=20
the late twenty-first century, we will have to select our body, our=20
personality, our environment--so many difficult decisions to make!=20
But don't worry--we'll have swarms of machines to guide us... Stroll=20
with you lover along a virtual Champs Elys=E9es, take a walk along a=20
virtual Cancun beach, mingle with animals in a simulated Mozambique=20
game reserve. Your whole relationship can be in Cyberland... With=20
this technology, you will be able to have almost any kind of=20
experience with just about anyone, real or imagined, at any time...=20
The nanobot swarms can instantly take on any form and emulate any=20
sort of appearance, intelligence, and personality that you or it=20
desires--the human form, say, if that's what turns you on...
To quote Max Beerbohm: "So this is utopia, is it? Well,/ I beg your=20
pardon, I thought it hell."
And who's to say it won't happen? Kurzweil, after all, was named=20
Inventor of the Year by MIT in 1988, and he and his sort are already=20
busy scanning brains, writing software that composes bad poetry and=20
cracks feeble jokes, and inventing neural implants to turn our minds=20
to software so that we can "reprogram" our feelings. If we cannot=20
stop such people, we at least have a solemn duty to do everything in=20
our power to mock, vex, and annoy them as they carry out their=20
ridiculous schemes. Don't you think, David?
=20
1/5
From: David Bennahum
To: Jim Holt
Subject: When the God Game Gets Real the Game Is Over
Posted: Tuesday, 3:54:00 PM PDT
Dear Jim:
I was surprised by your response--you still cling to the idea that=20
computers haven't added anything to the "sum of human happiness." And=20
you say that I responded to your "hard evidence" of economic=20
statistics with anecdotal evidence. Well, fine. I did. So? At the end=20
of the day, the point I was trying to make was that I do not care if=20
computers help, or do not help, General Motors build better widgets.=20
That measure for me has little to do with proving, or disproving, the=20
idea that computers have, or have not, added to the "sum of human=20
happiness." I can't speak for humanity; all I can do is speak to my=20
own experience, and that of the people I know around me.
I find the whole question strange--"happiness" is not about=20
technology, or media, or tools. It is an emotion triggered by the=20
mind, by our internal monologue, stimulated by events in the outside=20
world. For all I know, computers make some people wither in abject=20
horror, sort of what I would feel like were I to come upon a corpse=20
in my bed. I guess you sometimes feel that way around these machines.=20
I do not. There are moments, like last week when everyone was on=20
vacation, and I was forced to not work, that all I wanted to do was=20
play video games. I am particularly eager to play Myth II, a violent=20
strategy game involving a realistic 3D model of the world, with a=20
complete set of physics guiding the movement of all objects, from the=20
limbs of exploded dwarves, to the movement of scared chickens, all=20
compounded with realistic weather effects, like snow, rain, and wind.=20
=46or me computers are extensions of fantasy. They are vessels we can=20
go to for illusion, much as we can go to a book for the immersion of=20
a story.
I see very little difference between the joys of a well-crafted=20
computer game and the joys of a good book. They both evoke imaginary=20
worlds--except one does it with the language of words, and the other=20
does it with the language of visual objects, a kind of dynamic=20
pictograph, the gif as glyph. Which partly explains the seduction of=20
Kurtzweil and Gershenfeld's fantasies. Kurzweil understands that the=20
computer, as a modeling machine, can model reality, and that with the=20
right good fortune (for him, that is), the machine could eventually=20
manufacture authentic reality. His image of a "utility fog"--a set of=20
microscopic machines, acting in unison to produce objects atom by=20
atom on demand--is really a version of the "holodeck" on Star Trek.=20
The utility fog goes one step beyond modeling reality: It makes it.=20
But the link to that fantasy comes from computer games.
The desire I have to lose myself in the new version of Myth--the=20
weird way a good game seems to take you out of the body, and to=20
another place (much as a book does)--is a cousin, perhaps an=20
ancestor, to Kurtzweil's desire to lose himself in a utility fog, or=20
upload his brain into some sort of new container. But where we differ=20
is in the belief that this is either feasible, or worth doing. What=20
Kurtzweil reveals is how seductive computers can be, especially the=20
desire to use them to (re)model reality.
Kurtzweil's book has everything to do with computer games. Known as=20
"god-games," this is a branch of computer games where the player runs=20
a society (SimCity is a famous god-game). The Age of Spiritual=20
Machines is a literary god-game, with Kurtzweil laying out his=20
fantasy for the future. Sadly, these sorts of games tend to reveal=20
the same thing: The god is capricious, and the ultimate authority on=20
what's good or bad. Spiritual Stalinism. Kurtzweil's desire to=20
obliterate the old reminds me of turning on the "natural disaster"=20
button in SimCity, to see what happens when the tsunami crashes into=20
your carefully crafted metropolis. You ask whether Kurtzweil's=20
fantasy may happen. I can't imagine it will. At least not the way he=20
thinks.
We are far more likely to see the fusion of humans and machines=20
through medicine, not virtual reality, cyberspace, or computers.=20
Prosthetic limbs, enhanced cochlear implants, corrective lenses sewn=20
into the eyeballs, improved pacemakers--these lead to the=20
mechanically enhanced body (why not let the new lens pick up=20
infrared, along with the regular spectrum?). Oddly, genetic=20
engineering is missing from Kurtzweil's book. Which is very strange.=20
But then again, messy stuff like genetic engineering is too "body=20
based." Kurtzweil likes to think about the brain foremost, and=20
plugging computers right into our synapses has a higher nobility for=20
him. It's more "spiritual."
I am not worried about Kurtzweil's ideas--they've been written=20
before, as I mentioned in my previous message. Gershenfeld is far=20
more of a doer. He is a man with funding, a team of smart students,=20
and a plan that's not too ambitious. If you want to look at a far=20
more plausible future, albeit a less sexy one than Kurtzweil's,=20
Gershenfeld is the author to look to. We haven't talked much about=20
this book, but it deserves some attention. His fundamental premise=20
that the interface between us and computers must be rendered=20
physical, and that computers should be given the "senses" to "sense"=20
their surroundings, is right. It's right because it's useful. I=20
remember seeing a demonstration at the Media Lab of two people=20
exchanging business cards through the electric field in their skin as=20
they shook hands, and thinking: Yes, this makes sense. Does it make=20
sense to you?
/d