Emotional Investment

If imagination precedes discovery, then credit Isaac Asimov’s “I, Robot,” with its pulp fiction plotting and its Laws of Robotics (“A robot may not injure a human being or, through inaction, allow a human being to come to harm....”) for defining our notion of what a robot should be. Thirty years ago the most advanced versions were carts that moved a few centimeters, then paused for hours to recompute their position before moving again; last September a host of tiny independent machines burrowed nimbly into the wreckage of the World Trade Center in search of victims.

But these devices’ mental acuity has not kept pace with their physical nimbleness. Overall, the goal of creating a machine of truly discerning intellect has remained stubbornly out of reach, like the endlessly beckoning target in Zeno’s paradox. The World Trade Center robots could navigate inside dark dusty tunnels but could not distinguish between a rock and a human limb; that task had to be left to humans viewing video images transmitted from underground.

The goal of creating a robot “mind” remains seductive. But the quest has long been hampered by confusion over what it is the searchers are looking for. Is a “mind” something possessed by dog, or a frog? Or is it a higher-level process granted only to humans? Does it require language? Does it even, in fact, exist, or is it an illusion resulting from a meld of countless incredibly efficient but zombie-like responses to stimuli? Can it, in the end, be manufactured?

With “Flesh and Machines: How Robots Will Change Us,” Rodney Brooks steps into this intellectual snarl with both feet. Habitues of computing and artificial-intelligence conferences around the world know Brooks as the irrepressible Pied Piper of robotics, renowned for his thorough grasp of artificial intelligence, his chipper Australian accent and his breezy specifications for a remote, intelligent Mars exploratory vehicle--”fast, cheap, and out of control”--which became the title of a well-received documentary film profiling Brooks and three other obsessives from varied walks of life.

In his latest book (his first not directed at a technical audience), Brooks chronicles his attempts to coax animate behavior from a brood of aluminum-and-silicon creations, as well as his consequent battles with the AI establishment. He may be the director of the artificial intelligence lab at MIT, but he launched his career by turning the discipline’s basic tenets upside-down--proclaiming that AI was doomed to stagnation unless it shed its preoccupation with static computer programs and accepted that the one true path to machine intelligence was “embodiment,” that is, placing the intelligence inside a package that, like an animal, interacted dynamically with and learned from its environment. “[A]ll previous work in artificial intelligence,” he tells us cheerfully, “was misguided.”

Brooks is convinced that his brand of AI will eventually lead to a type of consciousness. “Being a machine does not disqualify something from having emotions,” he writes. “And by straightforward extension, does not prevent it from being conscious.” The implications of this argument are prodigious, not least the moral implications. Can we ethically justify treating our machines as slaves--or maltreating or discarding them unceremoniously--if they possess feelings, aspirations, thought? Brooks suggests it may be a long time before we have to confront that question in the flesh, as it were, but he believes the day will come.

But whether physics and engineering are capable of actually inventing consciousness, rather than its simulation, is the question at the heart of this book. “Flesh and Machines” is populated by skittering blinking robots with fanciful names like Shakey, Genghis, Cog and Kismet. The latter two, developed at Brooks’ lab, have even managed to evoke emotional responses from human interlocutors. Of Kismet, a robot head equipped with cameras in its faux-human eyeballs and microphones in its ears (the entire get-up resembles a creature from the movie “Gremlins”), Brooks contends that its ability to train its eyes on human speakers and engage in a simple programmed conversational give-and-take made it the world’s first “truly sociable” robot: “Kismet is alive. Or may as well be. People treat it that way.”

Yes, but. People respond emotionally to any number of artificial objects--the cartoon Bambi, the average plush toy, a Volkswagen. None of these is therefore rendered “sociable,” much less “alive.” One wonders, in fact, whether Brooks and his students didn’t start misleading themselves the moment they decided to give their mechanical contraptions names, like pets.

Brooks indirectly acknowledges this pitfall by retelling the story of the computer program ELIZA, which was developed by the AI pioneer Joseph Weizenbaum to simulate the circular conversational technique of a trained psychotherapist (“Eliza, I’m bored.” “Why do you think you are bored?”). Weizenbaum’s goal was to demonstrate how easily a computer could simulate a conversation and thus to prove the bankruptcy of efforts to reproduce human cognition, and he was appalled when he started receiving letters from people begging for private consultations with his artificial therapist.

In the end, Brooks stakes out a position on artificial consciousness--that it’s possible, even inevitable--which one tends to hear far more often from engineers than from biologists, who having more experience with the astonishing complexity and variety of the brain are less inclined to view it, as Brooks does, as just another computer. The neuroscientist Gerald Edelman puts his finger on the problem when he reminds AI researchers that machines such as digital computers don’t “think” in the human sense; what they do is perform logical operations on symbols, the true meaning of which they are unable to discern. It’s a long way from there to an artificial consciousness capable of viewing itself and the world as a complex, interactive meld of experience, remembrance and growth.

Brooks dismisses such objections as a species-oriented chauvinism, but this is unworthy. The drive to discover how the purely physical and chemical reactions inside our bodies give rise to our perceptions of a dynamic Technicolor world may well prove to be the most important scientific quest of our age. Likely as not, it won’t be solved until biologists and engineers agree they each have something to contribute to the effort, rather than insisting that they themselves are the owners of the one true key.