Superbrains born of silicon will change everything. Previously intractable problems in science, engineering, and medicine will be a snap. Robots will rapidly displace humans from factories and farms

At the moment, computers show no sign of intelligence. This is not surprising, because our present computers are less complex than the brain of an earthworm. But it seems to me that if very complicated chemical molecules can operate in humans to make them intelligent, then equally complicated electronic circuits can also make computers act in an intelligent way. -- Stephen W. Hawking, physicist, 1998

Intelligent computers are now considered as inevitable as Moore's Law--the 1965 dictum predicting the geometric growth of semiconductor power. The lawgiver himself agrees. ''Silicon intelligence is going to evolve to the point where it'll get hard to tell computers from human beings,'' says Gordon E. Moore, chairman emeritus of Intel Corp.

IBM's Gerd Binnig*
But computer intelligence won't stop there. Many scientists assert that machines will rapidly become far smarter than Albert Einstein and Hawking rolled into one. Just as humans can design computers with superior number-crunching capabilities, Hawking figures savvy machines will create still better computers. At least by mid-century, and probably much sooner, computers could have smarts way beyond our ken.

Silicon will even give birth to new kinds of life, predicts Robert E. Newnham, a materials scientist at Pennsylvania State University. And the advantages of this silicon life--chiefly immortality and unimaginable brainpower--could inspire scientists to forge composite human-silicon life forms ''with a common consciousness that transcends all living beings.''

A NIGHTMARE? These wild notions no longer come just from science-fiction writers. They're slowly creeping into mainstream science. And researchers are waking up to the implications of the monumental event that's coming within many of their lifetimes: our first contact with an alien intelligence.

The arrival of silicon life will transform civilization. All our science and art, even our concept of self, stems ultimately from what our senses tell us about the world. But beings that can see radio waves and listen to starlight, that can feel the vast empty spaces in atoms of steel, will have a very different perception of reality. What we learn from them could be more wondrous than all the discoveries made with microscopes, telescopes, X-ray machines, and other high-tech tools for amplifying our senses.

Some researchers fear super-brainy machines will be a science-fiction nightmare come true. Kevin Warwick, head of cybernetics research at Britain's University of Reading, is convinced that machines will subjugate humanity by 2050. And Hugo de Garis, head of a project to build silicon brains at Japan's Advanced Telecommunications Research Institute International (ATR), admits he is haunted by the prospect that his creations might ''swat me like a fly.''

Other researchers figure such beings would be too wise not to respect life in all its myriad forms. The idea of malevolent machines is based on the mistaken assumption that intelligent machines would behave pretty much like people, ''foibles and all,'' scoffs Igor Aleksander, head of neural systems engineering at London's Imperial College of Science, Technology & Medicine. But sexless creatures that know they are machines and can exist essentially forever wouldn't be driven to compete for territory and mates--two main sources of human inhumanity and maltreatment of lower life forms. So, if supersmart machines come to regard people as unfit company, perhaps they'll just build cylinders around themselves and blast into space. Some may do so anyhow, seeking new knowledge, since space travel will be a breeze for them.

BRAINS IN A BOX. Either way, the human brain has only a short time left as the smartest thing on earth. The speed and complexity of computers will continue to double every 18 months through 2012. By then the density of computer circuits will have jumped 1,000-fold, and the raw processing power of a human brain will fit into a shoe box. With luck, that milestone might come a lot sooner--perhaps as early as 2005, says John C. Carson, chief technology officer at Irvine Sensors Corp., a Silicon Valley chip company.

Beyond 2012, chips that exploit the quirky world of quantum mechanics promise far bigger leaps in complexity. Because such chips won't need wires, which now occupy most of the space on silicon, it won't take long to duplicate a human brain fully--not only its 100 billion neurons but also its trillions of synapses, or interconnections. This dense maze of interconnections is regarded as essential for intelligence to emerge. Hardware brains will get there by 2020, predicts Raymond C. Kurzweil, founder of Kurzweil Technologies Inc.

Then they'll soar way past human ''wetware.'' A billion human brains could soon be crammed into a cubic inch of quantum circuitry, Kurzweil says. And the size of artificial brains won't be constrained by the human skull. They could grow as big as trucks. De Garis of ATR even sees brains the size of satellites orbiting the earth.

Critics contend that no matter how big computers get, they can't become intelligent until we know how to emulate the brain's functions in software. Not so, retorts Inman Harvey, a mathematician turned roboticist at Britain's University of Sussex. By mimicking evolution, ''it's possible to create artificial brains without really understanding how they work,'' he says. In other words, they could evolve their own internal programming, just as human brains have.

ROBOTIC ROAD RAGE? These superbrains will change everything. Previously intractable problems in science, engineering, and medicine will be a snap. After 2025, Kurzweil says, robots will rapidly displace humans from factories and farms, and they'll provide basic human necessities to all people. Cars, planes, and trains will operate themselves, and the carnage on the highways will end in the 2030s.

Even the nature of human life itself will be changing by mid-century. Neural implants will expand human knowledge and thinking powers--and begin a transition to composite man-machine relationships that will gradually phase out the need for biological bodies. Swarms of microscopic robots will take up positions in the brain's sensory areas and create virtual-reality simulations that are impossible to distinguish from real reality. Communicating with family and friends won't require your physical presence. The best food you've ever eaten can be enjoyed time and again with different companions. And traveling to Mt. Fuji or the Louvre will be pointless, because your body won't be able to do or sense anything that can't be provided by in-brain simulations.

So, come 2099, Kurzweil figures only a very small group of people will still inhabit biological bodies. Most humans will have transferred their minds into electronic circuits--and attained immortality as a result (page 100).

Penn State's Newnham is sorry he won't have that opportunity, because he's already 70. ''I would like to live such a life,'' Newnham says wistfully. ''I would like to have the time to learn why life is, why we are here, why there is matter, and why the universe exists. I'd like to know those answers.''


To read a correction/clarification about this story, click here.

*Binnig is IBM's research director. Video interview by Otis Port (Needs G2 Player)

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

12: Machines Will Be Smarter Than We Are

ONLINE ORIGINAL: Half-Man, Half-Machine: The Mind of the Future

VIDEO Gerd Binnig, Director of Research, IBM
(Needs G2 Player)

Time is on the Cyborgs' side

The Next Phase of Evolution

Advanced Telecom Research Institute (ATR)

ATR's Brain Builder group

Swiss Federal Research Institute's Logic Systems Lab

University of Sussex evolutionary hardware and adaptive systems

Nick Bostrom, Philosophy Dept., London School of Economics

Hans Moravec's "Robot: Mere Machine to Transcendent Mind"

Symposium on Roger Penrose's "Shadows of the Mind"

Further Reading & Resources

CMU Artificial Intelligence Repository

Contemporary Philosophy of Mind Bibliography

MIT's AI Lab

University of Edinburgh's AI applications group

American Association for Artificial Intelligence

E-Mail to Business Week Online

Copyright 1999, Bloomberg L.P.
Terms of Use   Privacy Policy