Link zur Vorlesung!
About the Lecture
Two of the sharpest minds in the computing arena spar gamely, but neither scores a knockdown in one of the oldest debates around: whether machines may someday achieve consciousness. (NB: Viewers may wish to brush up on the work of computer pioneer Alan Turing and philosopher John Searle in preparation for this video.)
Ray Kurzweil confidently states that artificial intelligence will, in the not distant future, “master human intelligence.” He cites the “exponential power of growth in technology” that will enable both a minute, detailed understanding of the human brain, and the capacity for building a machine that can at least simulate original thought. The “frontier” such a machine must cross is emotional intelligence—“being funny, expressing loving sentiment…” And when this occurs, says Kurzweil, it’s not entirely clear that the entity will have achieved consciousness, since we have no “consciousness detector” to determine if it is capable of subjective experiences.
Acknowledging that his position will prove unpopular, David Gelernter launches his attack: “We won’t even be able to build super-intelligent zombies unless we approach the problem right.” This means admitting that a continuum of cognitive styles exists among humans. As for building a conscious machine, he sees no possibility of one emerging from even the most sophisticated software. “Consciousness means the presence of mental states strictly private with no visible functions or consequences. A conscious entity can call on a thought or memory merely to feel happy, be inspired, soothed, feel anger…” Software programs, by definition, can be separated out, peeled away and run in a logically identical way on any computing platform. How could such a program spontaneously give rise to “a new node of consciousness?”
Kurzweil concedes the difficulty of defining consciousness, but does not want to wish away the concept, since it serves as the basis for our moral and ethical systems. He maintains his argument that reverse engineering of the human brain will enable machines that can act with a level of complexity, from which somehow consciousness will emerge.
Gelernter replies that believing this “seems a completely arbitrary claim. Anything might be true, but I don’t see what makes the claim plausible.” Ultimately, he says, Kurzweil must explain objectively and scientifically what consciousness is -- “how it’s created and got there.” Kurzweil stakes his claim on our future capacity to model digitally the actions of billions of neurons and neurotransmitters, which in humans somehow give rise to consciousness. Gelernter believes such a machine might simulate mental states, but not actually pass muster as a conscious entity. Ultimately, he questions the desirability of building such computers: “We might reach the state some day when we prefer the company of a robot from Walmart to our next-door neighbor or roommates.”