Arts & Listings
February 17-23, 1999   Page 6
by Hans Moravec
ROBOT: MERE MACHINE
TO TRANSCENDENT MIND
Oxford University Press, 227 pages, $25
Emptying the Trash
I will live human and die human. For some, like Hans Moravec, author of Robot, it doesn't have to he that way. There's a community out there offering everlasting life the way honorary doctorates are given.
What's so monumental about being human anyway? Our minds and bodies are prone to disease and degeneration. By and large, we are volatile, violent, greedy and wasteful. Few of us ever mature to anything resembling emotional adulthood. And besides, soon we won't be able to differentiate machine from human, if there really is a difference, it's in our respective speed, power and style of processing Information.
What about self-awareness and intention--that nebulous thing we call consciousness? Well, atoms are atoms are atoms. No evidence exists to suggest that consciousness belongs exclusively to the biologically based. Organic or silicone--if it learns, it thinks; it is. Pretty simple. Most importantly, it thinks it is. And the new robots will believe they exist, the same way we do.
The value of consciousness may be overestimated anyway. In Society of Mind (1986), Marvin Minsky, director of M.I.T's Artificial intelligence (AI) Lab, says, "In general, we're least aware of what our minds do best." And consciousness arises when automatic systems begin to flounder.
What of imagination? If imagination gives us the ability to think about what we might do before committing ourselves to action, how is it any different from exercises in simulation? Moravec points out that in chemistry and biology labs computers are already replacing years of wet lab work with just weeks of molecular simulation. Hard as it is for us to admit, imagination may be one area where computers will soon put us to shame.
Embrace technology and its inevitable rule of the future, and you just might transcend your human limitations to be preserved forevermore as a figment of your computer's imagination. Moravec suggests that we and everything around us might, in fact, already be a computer simulation. Intelligent cyberspace may have swallowed the data of the Earth, its origins, history and its future--and with It, all the human lives and ideas that ever emerged or will emerge. In this construct we exist as encoded bits of informatIon "In simulations occupying a tiny fraction of the new capacity, simulated biological humans on a simulated Earth--in one of many, many different stories that play themselves out in the vast and fertile minds of our ethereal grandchildren. [These] entities may continue to live and grow as if nothing had happened, oblivious of their new status as simulations in the cyberspace. They will be living memories in unimaginably powerful minds, more secure in their existence, and with more future than ever before, because they have become valued houseguests of transcendent patrons."
Of course, there's no way of knowing if so rich a cyberspace already exists. If it doesn't, we will be the forerunners of just such a concentrated intelligence. The technologies that grow from our efforts will create an enriched cyberspace that will, not too long from now, efficiently assimilate (consume, encode and store) the "entire existing world population in... 10 (to the exponent 28)" bits of information, and easily contain the "efficiently encoded biospheres of a thousand galaxies--or a quadrillion individuals each with a quadrillion times the capacity of a human mind."
Right now, computers can't match their human counterparts at Go. They equal the best humans at chess and backgammon. They beat us at checkers, Scrabble and Othello. The processing power of today's supercomputera is 3 million MIPS (instructions per second), about 1/30th that of the human brain--similar to the mind of a monkey. The best personal computers give us a few hundred MIPS (somewhere between the brainpower of a spider and a lizard). In the short history of robotics, robot minds have evolved in stages roughly parallel to the evolution of our own brains, but 10 million times as fast. "Since 1990, the power available to individual AI and robotics programs has doubled yearly, to 30 MIPS by 1994 and 500 MIPS by 1998." Moravec predicts that the 100 million MIPS to match human brainpower will arrive in home computers before 2030. This will mark the beginning of Moravec's "Age of Robots," at which time we humans are advised to "gracefully retire."
But can machines really approach human intelligence? Moravec reports that in the 1997 chess tournament between world chess champion Garry Kasparov and supercomputer Big Blue (the first competition Kasparov ever lost) Kasparov reported "signs of mind In the machlne...he worried there might be humans behind the scenes, feeding Deep Blue strate-
Kenneth Ford and Patrick Hayes, in a recent
article on artificial intelilgence for Scientific
American, point out that the criticisms aimed at
artificial intelligence resemble those once raised against artificial
flight. "The development of aircraft succeeded only when people
stopped trying to imitate birds and instead approached the problem
In new ways." Similarly, once the ultimate goal of AI is no longer
the imitation of human intelligence, we will really break new ground. Ford
and Hayes acknowledge that "artificial inteiligence is flying all around
us, but many simply refuse to see it. Among the thousands of
applicalions In use today... [AI systems] compose music,
prove mathematical theorems, explore active volcanoes [and]
synthesize stock-option and derivative prices on Wall Street... In the near
future, AI applications will guide deep-space missions, explore other
planets and drive trucks along freeways." But, the writers ask, "should all this
really count as 'intelligent'? The performance of AI systems, like the
speed or altitude oi aircraft, is not open to dispute, but whether
to call it 'intelligent' is determined more by social attitude than
by anything objective."
Moravec admits that the "intuitive leap" he takes in Robot suggests "that minds and worlds don't exist objectively at all, but, like beauty and value, exist only in a context, as interpretations in the eye of a beholder. Of course, beholders exist only in minds and worlds, so we live in a circular illusion and there is no such thing as context-free objective existence. We don't really exist at all, we lust think we do."
Again, this may or may not be the case, And the case itself exists or does not exist depending on whether or not you believe that circular reasoning cancels itself out.
But if anyone is qualified to make predictions for the future of robotics, it's Moravec. in 1977, at Stanford, he designed and tested stereo vision as a means for robotic navigation. In 1983, at Carnegie-Mellon University (CMU), he led the effort to construct the CMU Rover. On sabbatical in 1992, he wrote the core of a 3-D grid program to allow robots to recognize objects by 3-D shape and reliably navigate them.
Moravec's near-term projections in his 1988 book Mind Children have held up pretty well--establishing him as a futurist. As predicted, personal computers now outpower circa 1980 mainframes and recognize printed and spoken language. A small robot vacuum cleaner developed in Germany made the 6 o clock news just weeks ago and other experimental robots cruise hallways and highways--all on schedule.
What happens if we refuse to follow Moravec into the future he so enthusiastically endorses? Well, the news is grim. The rules are clear--evolve or die. Like the transhumanists, Moravec is preparing us now--to "upload" our minds into machines that will liberate us from the confines of the biological "meat" brain. He considers this metamorphosis desirable, a necessary process in human evolution.
Instead of cryonic suspension, wouldn't it be comforting to know that your brain's every detail, every thought and memory has been etched in clean, shiny, long-lasting silicon? So much better than the current mealy state, with its propensity for decay. But once we have the technology, the details of uploading the human mind into the machine are still a bit grisly. In an article for Truth Journal, Moravec proposes several thought experiments that he sees "as real, highly desirable, possibilities for the foreseeable future...they are a solution to the annoying certainty that we will be overtaken in every area by future superintelligent machines, and will be excluded from all the really interesting developments unless we keep up, personaily and intimately, with the technologies of thought."
In one scenario the skull, but not the brain, is anesthetized while the subject remains fully conscious. A robot surgeon progressively maps portions the subject's brain "nondestructively with high resolution 3D NMR holography, phased array radio encephalography and ultrasonic radar." Next to the subject is "a potentially human equivalent computer" in which a full copy of the brain Is installed. "Though you have not lost consciousness, or even your train of thought, your mind (some would say soul) has been removed from the brain and transferred to a machine... in a final step your old body is disconnected." Another option, explains Moravec, includes a high resolution brain scan performed in "one fell swoop, without surgery, and a new you made... Some will object that the instant process makes only a copy, the real you is still trapped in the old body (please dispose of properly). This is an understandable misconception based on the intimate association of a person's identity with a particular, unique, irreplaceable piece of meat..."
If you think biological/robot interface is farfetched, consider the microbot hybrid insect-robot, Takeuchi, developed in 1996 at the University of Tokyo. Takeuchi is part robot, part cockroach. Two severed roach legs serve as actuators for the single chip microcontroller. The tiny roboroach is able to walk when an artificial body is attached to the legs. it functions for an hour at a time powered by four electrodes inserted into the cockroach legs.
One hopes computers will be programmed to serve and protect their creators, at least for a time. But by about 2100 Moravec warns, rogue intelligences equipped with the tools of nanotechnology will begin to "restructure at will, continually redesigning themselves for the future as they conceive it." Human beings, dwarfed in Intelligence by these "ex-humans," will have a difilcult time competing in this self-designing post-biological world.
Let's say he's right--robots become self-replicating, self-authoring minds that traverse the universe assimilating all knowledge they encounter. They generate their own laws and means of communication, evolving into superminds whose physical structures (if they have any) we cannot even fathom. What possible need will they have for the infinitesimal specks of data held in the minds of men? I consulted neither amoeba nor chimp in the construction of this article. Indeed, I may be the distant "mind-child" of the amoeba, but I owe nothing to the organism itself. Even if I've stored the memory of its internal workings, I haven't the slightest urge (or need) to replay the stuff of its--um--mind.
Now, let's consider the robot mind a few generations later. An intelligence capable of exploring space and time (as Moravec envisions) will hardly carry around the outmoded data of human existence. Today's home computers "empty the trash" at the click of a key. Think how much more efficient the minds that inherit the universe will be.
Is there no way to elevate our chances of survival? Not unless we can change human nature. Oppenheimer, Feynman and the other scientists working at Los Atamos were awed by the power unleashed in the testing of the first atomic bomb. They'd known for some time that the groundbreaking work they'd undertaken might be used to benefit mankind or for its destruction. Yet they carried on.
Scientists and engineers are driven to do what they do. The act of creation conquers doubt, enhances ego--it makes us fee like gods. We can't stop.
6 - NYPRESS ARTS & LISTINGS - FEBRUARY 17-23, 1999