Artificial
Intelligence I’m sure that Hans Moravec is at least as sane
as I am, but he certainly brought to mind the classic mad scientist as we sat in
his fifth-floor office at Carnegie-Mellon University on a dark and stormy night.
It was nearly midnight, and he mixed for each of us a bowl of chocolate milk and
Cheerios, with slices of banana piled on top. Then, with
banana-slicing knife in hand, Moravec, the senior research scientist at Carnegie
Mellon’s Mobile Robot Laboratory, outlined for me how he could create a robotic
immortality for Everyman, a deathless universe in which life would go on
forever. By creating computer copies of our minds and transferring, or
downloading, this program into robotic bodies, Moravec explained, humans could
survive for centuries. "You are in an operating room. A robot
brain surgeon is in attendance ... Your skull but not your brain is anesthetized
(麻醉). You are fully conscious. The surgeon opens your braincase and peers
inside." This is how Moravec described the process in a paper he wrote called
"Robots That Rove". The robotic surgeon’s attention is directed at a small clump
of about one hundred neurons somewhere near the surface. Using high-resolution
3-D nuclear-magnetic-resonance holography, phased-array radio encephalography,
and ultrasonic radar, the surgeon determines the three-dimensional structure and
chemical makeup of that neural clump. It writes a program that models the
behavior of the clump and starts it running on a small portion of the computer
sitting next to you. That computer sitting next to you in the
operating room would in effect be your new brain. As each area of your brain was
analyzed and simulated, the accuracy of the simulation would be tested as you
pressed a button to shift between the area of the brain just copied and the
simulation. When you couldn’t tell the difference between the original and the
copy, the surgeon would transfer the simulation of your brain into the new,
computerized one and repeat the process on the next area of your biological
brain. "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," Moravec said, "In a final step your old
body is disconnected. The computer is installed in a shiny new one, in the
style, color, and material of your choice." As we sat around
Moravec’s office I asked what would become of the original human body after the
downloading. "You just don’t bother waking it up again if the copying went
successfully." he said. "It’s so messy. Humans have got so many problems that
you might just want to leave it retired. You don’t take your Junker car out if
you’ve got a new one." Moravec’s idea is the ultimate in life
insurance. Once one copy of the brain’s contents has been made, it will be easy
to make multiple backup copies, and these could be stashed in hiding places
around the world, allowing you to embark on any sort of adventure without having
to worry about aging or death. As decades pass into centuries you could travel
the globe and then the solar system and beyond--always keeping an eye out for
the latest in robotic bodies into which you could transfer your computer
mind. If living forever weren’t enough, you could live forever
several times over by activating some of your backup copies and sending
different versions of yourself out to see the world. "You could have parallel
experiences and merge the memories later," Moravec explained. In
the weeks and months that followed my stay at Carnegie-Mellon, I was intrigued
by how many researchers seemed to believe downloading would come to pass. The
only point of disagreement was when--certainly a big consideration to those of
us still knocking around in mortal bodies. Although some of the researchers I
spoke with at Carnegie-Mellon, MIT, and Stanford and in Japan thought that
downloading was still generations away, there were others who believed achieving
robotic immortality was imminent and seemed driven by private passions never to
die. The significance of the door Moravec is trying to open is
not lost on others. Olin Shivers, a Carnegie-Mellon graduate student who works
closely with Moravec as well as with Allen Newell, one of the founding fathers
of artificial intelligence, told me, "Moravec wants to design a creature, and my
professor Newell wants to design a creature. We are all, in a sense, trying to
play God." At MIT I was surprised to find Moravec’s concept of
downloading given consideration by Marvin Minsky, Donner Professor of Science
and another father of artificial intelligence. Minsky is trying to learn how the
billions of brain cells work together to allow a person to think and remember.
If he succeeds, it will be a big step toward figuring out how to join perhaps
billions of computer circuits together to allow a computer to receive the entire
contents of the human mind. "If a person is like a machine, once
you get a wiring diagram of how he works, you can make copies," Minsky told
me. Although Minsky doesn’t think he’ll live long enough to
download (he’s fifty-seven now), he would consider it. "I think it would be a
great thing to do." he said, "I’ve spent a long time learning things, and I’d
hate to see it all go away." Minsky also said he would have no
qualms about waving good-bye to his human body and taking up residence within a
robot. "Why not avoid getting sick and things like that" he asked. "It’s hard
to see anything against it. I think people will get fed up with bodies after a
while. Then you’ll have another population problem: You’ll have all the people
of the past, as well as the new ones." Another believer is Danny
Hillis, one of Minsky’s Ph. D students and the founding Scientist of Thinking
Machines, a Cambridge-based company that is trying to create the kind of
computer that might someday receive the contents of a brain. During my research,
several computer scientists would point to Hillis’s connection machine as an
example of a new order of computer architecture, one that’s comparable to the
human brain. (Hillis’s connection machine doesn’t have one large central
processing unit as other computers do but a network of 64,000 small
units--roughly analogous in concept, if not in size, to the brain’s network of
40 billion neuronal processing units. ) "I’ve added up the
things 1 want to do in my life, and it’s about fifteen hundred years’ worth of
stuff," Hillis, now twenty-eight, told me one day as we stood out on the
sixth-floor sundeck of the Thinking Machines building. "I enjoy having a body as
much as anyone else does, but if it’s a choice between downloading into a
computer--even one that’s stuck in a room someplace-- and still being able to
think versus just dying, I would certainly take that opportunity to
think." Gerald J. Sussman, a thirty-six-year-old MIT professor
and a computer hacker of historic proportions, expressed similar sentiments.
"Everyone would like to be immortal. I don’t think the time is quite right, but
it’s close. I’m afraid, unfortunately, that I’m in the last generation to
die." "Do you really think that we’re that close" I
asked. "Yes," he answered, which reminded me of something
Moravec had written not too long ago: "We are on a threshold of a change in the
universe comparable to the transition from nonlife to life." In Moravec’s mind, when a robotic immortality is created, life will go on forever.