Future path of artificial intelligence

ROGER FRANKLIN reports on film fantasy and future robotic reality.

NEW YORK - Unlike Steven Spielberg, who infuses AI: Artificial Intelligence with its bleak luminosity, the people who built Technology Square on the Massachusetts Institute of Technology campus in Cambridge, just outside Boston, had a very different future in mind.

The world of robots, computers and sentient silicon that was to be brought into being within the complex's four bluff buildings would be a realm of right-angled rectitude, a four-square Utopia designed by engineers. Only such a sterile creed could have inspired the grass-and-concrete checkerboard that is the complex's central square.

A visitor navigating the wide and empty space on his way to the office of Professor Rodney Brooks, the world's most controversial roboticist, plots his ant-like progress across a gigantic sheet of graph paper.

Tall, thin, dark-tinted slits of windows lend Brooks' workplace a fortress air, and the impression that you have entered a citadel of science grows as you cross a lobby of austere marble. A security guard examines all credentials. He is armed, because, well, this is America, and also because much of the research being carried out in the building is funded by the Pentagon. The elevator rises to Brooks' sanctum, where a receptionist establishes the visitor's bona fides for a second time before the door to MIT's Artificial Intelligence Laboratory springs open with a muted buzz.

It is only then, over those final few metres to Brooks' office, when it hits you that it is Spielberg, rather than the builders of Technology Square, who has pegged the future right. In A I, a fractured fairytale which opens in New Zealand tomorrow about a robotic Pinocchio that was originally conceived by the late Stanley Kubrick, science hasn't made the world all neat and tidy. What it has achieved is the exact opposite: Technology has magnified mankind's flaws and made a basket case of the planet, which is dying a hot and sweaty death from global warming.

Far from redeeming us, science has made it possible - almost irresistible - to indulge all our age-old lusts and self-deceptions. Every need has its machine. There are sexually expert cyborgs to make humanity writhe with pleasure on its deathbed, and robot slaves to do the heavy lifting. And then there is David, a prototype boy "mecca" played by The Sixth Sense's Haley Joel Osment, whose appeal is that he can be programmed to serve as a surrogate and substitute for a desperate mother's sickly son. In the Spielberg/Kubrick's faithless universe, even a mother's heart can be turned on and off as changing emotions and shifting loyalties demand.

When the movie's human child recovers, the robot boy is cast aside and his journey to an uncertain fate begins. As American critics have noted, this movie is a far, far cry from ET's fuzzy warmth. In that earlier film, Spielberg lionised children for their innocent courage and unqualified goodness. In A I, he - or maybe it's really Kubrick's ghost - says that, in humans and robots alike, everything springs from the etched circuits of programmed emotion.

What tips the future to Spielberg and silences Technology Square's bricks-and-mortar rendition of science uber alles is the maze of chaotic computations on the chalkboard outside Brooks' office. At the board's centre, framed by that tangled halo of mathematical squiggles, is a sketch of a simple cube. Each of the cube's dimensions - length, breadth and height - represents a single concept: Joy. Sadness. Self. If something like AI's child robot is ever built, the owner's manual will likely explain the machine's ability to replicate "love" with just such a drawing. Brooks' blackboard is the confirmation that, when the future arrives, sheer clinical logic won't be an end in itself. What science has wrought will be placed at the service of emotion, or some artificial facsimile very much like it.

How so? According to Brooks, who worked as a consultant on A I and hosted its premiere in his lab's theatre, it works this way: Pick a point inside the cube and assume that it represents a person's self-awareness at any given moment. Now move it all the way to "joy," and the mood of a creature - robot or boy - will change to one of ecstasy. Slide it upwards and away from "self" and ecstasy becomes reckless abandon. Or here is another way to think about it: A fellow takes a few drinks and becomes overly cheerful. He downs a few more, inhibition recedes, and he wakes up the next morning with a headache, the barmaid, and a guilty conscience because his emotional parameters have returned to their normal settings.

Brooks has been fascinated by the idea of robots and artificial intelligence since his undergraduate days. Spielberg's lost doll of a little boy doll remains a dream - but not a distant one, of that Brooks is certain. Nor are the Robotic Age's dilemmas and temptations so far away, either.

The reason for Brooks' confidence holds pride of place in the Artificial Intelligence Laboratory's main workshop. It is a machine called Kismet, a cute little guy with moving eyes, ears that wiggle in response to stimuli and cartoon lips that are really lengths of red surgical plastic. Enter the room, and Kismet "wakes" to greet you with a cheery, expectant smile.

Take things slowly, smile a lot, avoid jerky movements, and Kismet is "happy." If you make a cheerful face, he'll give you a radiant smile in return. Overload his optical circuits and sensors with too much stimulation, however, and he grows flustered. Just like a startled infant, his features register emotional "pain." Then he is apt to switch himself off with obvious "disgust" and take an exhausted "nap" to regain emotional equilibrium. The reactions are so uncanny, so darn cute, it's impossible not to feel a strong and immediate empathy. This is precisely what Brooks and his associate, Cynthia Breazeal, intended when they assembled Kismet's engaging features: they wanted the robot to entice human interaction in order to give its circuitry more opportunities to "learn" the appropriate responses. The gizmo could easily fool an observer into believing there is an independent, native intelligence inside what is, as Brooks explains, a basic architecture of off-the-shelf chips. Other artificial intelligence researchers have taken the opposite track, writing millions of lines of code that have so far failed to replicate the mechanics of true human thought.

Deep Blue, the IBM computer that beat Gary Kasparov, is the result of that approach. After marshalling all its lines of software code, it can defeat a chess grandmaster by blasting through millions of possible moves and selecting the sequence most likely to succeed. Ask it to boil an egg and it will be a long time between meals.

Despite Kismet's slick tricks, Brooks acknowledges that he has also failed to create a form of truly autonomous intelligence. Kismet succeeds to the extent he does because, like the mother in AI, we humans are eager to be gulled if there is an emotional payoff. "We compulsively project our own feelings and emotions. As a species, we anthropomorphise everything."

The acid test will come, he predicts with a prophet's certainty, when a Kismet-style robot produces responses that are so perfect, the temptation to forget it isn't human will be irresistible. Will that be a true example of artificial intelligence, or will it be intelligence artificially projected by the observer?

While Spielberg's boy-toy robot looks and walks and talks like a real boy, he's no better than a cute package of hard-wired reactions - just like Kismet. While David can produce an impersonation of love, he cannot recognise the absence of the real thing in others. And that is another reason why AI is such a Spielberg oddity. Although he permits his audience the faintest glimmer of a chance to smile at the end, he keeps it logically bleak. As Brooks would note, it must have taken a supreme effort - a human effort - to circumvent Hollywood's most shameless sentimentalist's normal programming.