The soul of the human machine
This summer, Stephen Hawking and others warned that artificially intelligent machines could turn on us and destroy humankind. In Hollywood and Silicon Valley, there is talk of singularity, that critical juncture when self-instructing machines exceed our mental capacities. And, from Lisbon, there’s recent news of an AI program that answers SAT geometry problems. Face it — the machines are winning.
If these devices can think so well, then perhaps it also follows that we are machines, and not very good ones. Numerous AI researchers, neuroscientists, and cognitive psychologists now assume that brains are computational devices. Last month, in a move thick with symbolic significance, the scientist who directed the nation’s efforts to understand diseases of the brain and mind took a job at Google.
Well, rest assured: Human beings have never been machines. Despite centuries of eager analogizing, this equation has never added up.
Mechanical philosophy was first proposed in the late 16th century. Fifty years later, René Descartes declared that animals as well as human bodies were machines. The beauty of this conception lay in its power and its failings. For as Descartes’ friend the French friar Marin Mersenne knew, human cognitive faculties, those everyday acts of reasoning and willing, could never be explained by passive mechanistic action. Devout Western scientists could approach the natural world reductively and mechanistically, assured that their inner world required something else entirely: an active, immaterial soul.
This division between the machine and the soul immediately created insurmountable problems. Many critics sneered at the idea that wily foxes were tick-tocking contraptions. Skeptics asked how an immaterial thinking thing could possibly direct a material body. And the search for the soul’s seat in the brain yielded so many seats that one could host a dinner party.
It was inevitable that someone would try to untie these knots. That job fell to a French military doctor, Julien Offray de La Mettrie. While on the battlefield, he fell into a state of febrile visions that made him later conclude that consciousness was a mechanically created dream. In 1747, he published “Man-Machine,’’ in which he declared that conscious choice was a mirage. Mankind exercised free will the way that the hand on a watch “chose” to inch forward. La Mettrie is the unacknowledged forefather of our own eliminative reductionists in neuroscience and AI. He was the first modern thinker who dared to publicly propose that consciousness, reason, creativity, and free will were illusions, the effect — never the cause — of mechanisms in our heads.
Since that time, numerous mechanical devices have proved to be irresistible metaphors for scientists of the brain and mind, but the development of digital computers promised to be more than just an analogy. Alan Turing developed his “imitation game” to determine whether a digital machine could literally equal human thought. Today, some six decades later, the results of these efforts are in.
Digital computers have created astonishing, life-changing tools, but when it comes to their ability to think and act and communicate like a human mind and brain, they have been a bust. In a 2013 lecture, physicist Freeman Dyson surveyed 60 years of efforts and declared that the failure of AI in this regard was complete. His conclusion was that the brain simply was not a digital device.
Such scientific views go against the Apple-Google-Facebook zeitgeist. Today, fast, small computers talk to us, guide us, and answer our questions. As these devices continue to progress, it will seem like common sense to consider them not just gadgets, but little minds. That equation will severely limit our understanding of the biological, living, morphing, dynamically alive brain/mind. Mechanical philosophy can offer no clue to active mental functions — the capacity for “top-down regulation” that makes our minds so extraordinary.
Machines possess no capacity to will, create, and want. From inside the computational framework, powers like these can only be bracketed or dismissed. If widely accepted, the moral and political implications of such dismissals would be grave. What becomes of democracy, individual liberty, and the right to pursue happiness, if computer-man has no capacities for free choice and is algorithm-driven?
Scientific researchers require guiding analogies and metaphors to help them simplify the massive complexities and mysteries of the mind and brain. Loose, reductive equations may be pragmatically necessary, but when overstated they can prove blinding and at times dangerous. The idea that human brains are digital computers is just that.
George Makari is a professor of psychiatry at Weill Cornell Medical College. His forthcoming book is “Soul Machine: The Invention of the Modern Mind.’’