...So in terms of the debate topic, I thought it was interesting that there's an assumption in the title that we will build superintelligent machines, we'll build superintelligent machines that are conscious or not conscious. And it brings up the issue of consciousness, and I want to focus on that for a moment, because I think we can define consciousness in two ways. We can define apparent consciousness, which is an entity that appears to be conscious – and I believe, in fact, you have to be apparently conscious to pass the Turing test, which means you really need a command of human emotion. Because if you're just very good at doing mathematical theorems and making stock market investments and so on, you're not going to pass the Turing test. And in fact, we have machines that do a pretty good job with those things. Mastering human emotion and human language is really key to the Turing test, which has held up as our exemplary assessment of whether or not a non-biological intelligence has achieved human levels of intelligence.

And that will require a machine to master human emotion, which in my view is really the cutting edge of human intelligence. That's the most intelligent thing we do. Being funny, expressing a loving sentiment – these are very complex behaviors. And we have characters in video games that can try to do these things, but they're not very convincing. They don't have the complex, subtle cues that we associate with those emotions. They don't really have emotional intelligence. But emotional intelligence is not some sideshow to human intelligence. It's really the cutting edge. And as we build machines that can interact with us better and really master human intelligence, that's going to be the frontier. And in the ten minutes, I'll try to make the case that we will achieve that. I think that's more of a 45-minute argument, but I'll try to summarize my views on that.

I will say that the community, AI community and myself, have gotten closer in our assessments of when that will be feasible. There was a conference on my 1999 book, Spiritual Machines, at Stanford, and there were AI experts. And the consensus then – my feeling then was we would see it in 2029. The consensus in the AI community was, oh, it's going to – it's very complicated, it's going to take hundreds of years, if we can ever do it. I gave a presentation – I think you were there, Rodney, as well, at AI50, on the 50^th anniversary of the Dartmouth Conference that gave AI its name in 1956. And we had these instant polling devices, and they asked ten different ways when a machine would pass the Turing test – when will we know enough about the brain, when will we have sophisticated enough software, when'll a computer actually pass the Turing test. They got the same answer – it was basically the same question, and they got the same answer. And of course it was a bell curve, but the consensus was 50 years, which, at least if you think logarithmically, as I do, that's not that different from 25 years.

So I haven't changed my position, but the AI community is getting closer to my view. And I'll try to explain why I think that's the case. It's because of the exponential power of growth in information technology, which will affect hardware, but also will affect our understanding of the human brain, which is at least one source of getting the software of intelligence.

The other definition of consciousness is subjectivity. Consciousness is a synonym for subjectivity and really having subjective experience, not just an entity that appears to have subjective experience. And fundamentally – and I'll try to make this point more fully in my ten-minute presentation – that's not a scientific concept. There's no consciousness detector we can imagine creating, that you'd slide an entity in – green light goes on, OK, this one's conscious, no, this one's not conscious – that doesn't have some philosophical assumptions built into it ...rotransmitters?

Ray Kurzweil, is a noted inventor and technologist. He is the author of many books, most recently, "The Singularity Is Near: When Humans Transcend Biology".

First, I'd like to say thanks for inviting me. My guess is that the position I'm representing – the anti-cognitivist position, broadly speaking – is not the overwhelming favorite at this particular site. But I appreciate your willingness to listen to unpopular opinions, and I'll try to make the most of it by being as unpopular as I can. (laughter)

First, it seems to me we won't even be able to build superintelligent zombies unless we attack the problem right, and I'm not sure we're doing that. I'm pretty sure we're not. We need to understand, it seems to me, in model thought as a whole the cognitive continuum. Not merely one or a discrete handful of cognitive styles, the mind supports a continuum or spectrum of thought styles reaching from focused analytical thought at one extreme, associated with alertness or wide-awakeness, toward steadily less-focused thought, in which our tendency to free-associate increases. Finally, at the other extreme, that tendency overwhelms everything else and we fall asleep.

So the spectrum reaches from focused analysis to unfocused continuous free association and the edge of sleep. As we move down-spectrum towards free association, naturally our tendency to think analogically increases. As we move down-spectrum, emotion becomes more important. I have to strongly agree with Ray on the importance of emotion. We speak of being coldly logical on the one hand, but dreaming on the other is an emotional experience. Is it possible to simulate the cognitive continuum in software? I don't see why not. But only if we try.

Will we ever be able to build a conscious machine? Maybe, but building one out of software seems to me virtually impossible. First, of course, we have to say what conscious means. For my purpose, consciousness means a subjectivity. And Ray's – and consciousness means the presence of mental states that are strictly private, with no visible functions or consequences. A conscious entity can call up some thought or memory merely to feel happy, to enjoy the memory, be inspired or soothed or angered by the thought, get a rush of adrenaline from the thought. And the outside world needn't see any evidence of all that this act of thought or remembering is taking place.

Now, the reason I believe consciousness will never be built out of software is that where software is executing, by definition we can separate out, peel off a portable layer that can run in a logically identical way on any computing platform – for example, on a human mind. I know what it's like to be a computer executing software, because I can execute that separable, portable set of instructions just as an electronic digital computer can and with the same logical effect. If you believe that you can build consciousness out of software, you believe that when you execute the right sort of program, a new node of consciousness gets created. But I can imagine executing any program without ever causing a new node of consciousness to leap into being. Here I am evaluating expressions, loops, and conditionals. I can see this kind of activity producing powerful unconscious intelligence, but I can't see it creating a new node of consciousness. I don't even see where that new node would be – floating in the air someplace, I guess.

And of course, there's no logical difference between my executing the program and the computer's doing it. Notice that this is not true of the brain. I do not know what it's like to be a brain whose neurons are firing, because there is no separable, portable layer that I can slip into when we're dealing with the brain. The mind cannot be ported to any other platform or even to another instance of the same platform. I know what it's like to be an active computer in a certain abstract sense. I don't know what it's like to be an active brain, and I can't make those same statements about the brain's creating or not creating a new node of consciousness.

Sometimes people describe spirituality – to move finally to the last topic – as a feeling of oneness with the universe or a universal flow through the mind, a particular mode of thought and style of thought. In principle, you could get a computer to do that. But people who strike me as spiritual describe spirituality as a physical need or want. My soul thirsteth for God, for the living God, as the Book of Psalm says. Can we build a robot with a physical need for a non-physical thing? Maybe, but don't count on it. And forget software.

Is it desirable to build intelligent, conscious computers, finally? I think it's desirable to learn as much as we can about every part of the human being, but assembling a complete conscious artificial human is a different project. We might easily reach a state someday where we prefer the company of a robot from Wal-Mart to our next door neighbors or roommates or whatever, but it's sad that in a world where we tend to view such a large proportion of our fellow human beings as useless, we're so hot to build new ones. (laughter)

In a Western world that no longer cares to have children at the replacement rate, we can't wait to make artificial humans. Believe it or not, if we want more complete, fully functional people, we could have them right now, all natural ones. Consult me afterwards, and I'll let you know how it's done. (laughter)

David Gelernter, author of Americanism: The Fourth Great Western Religion, just published by Doubleday, is a professor of computer science at Yale and a National fellow at the American Enterprise Institute.