Once they get complex enough to be called androids and have what we call thoughts, we will not be in complete control of their mental activity in any case. Nor will they, of course.

Prof. David Ritz Finkelstein. School of physics and the Quantum Relativity Group, Georgia Institute of Technology, Atlanta, USA.

Yes, I think they will.

David Chalmers. Physicist, and author of The Conscious Mind: In Search of a Fundamental Theory.

Of course they will. They'll also dream of biological sheep, and humans, and whatever else pops into their minds. The term 'androids' is ill-defined, but I’m assuming it means any artificial, human-like robot. In the future, some of these constructions will have very human-like brains, because their brains will be built, connection for connection, from the patterns of human patients at the end of their biological lifespan. The android’s artificial brain will be functionally equivalent to the human's at the time his mind is 'uploaded.' So if the human had a propensity to dream about sheep, he will continue to do so as an android. But given the odd (to us) nature of his body at that point, it will no doubt strike him as interesting to consider what other animals might be converted from proteins to silicon. And if he ponders this while drifting off to sleep, then yes – electric sheep will jump in his dreams.

Joseph J. Strout. Department of neuroscience, University of California, San Diego, USA.

Essentially, only androids could ever know the answer to your question! Even if we one day have a much better idea of what it is that makes humans conscious (and I’m sure we will), we can only ever use this information to ask whether androids are conscious in the same way as humans are – the features of our brain which cause us to have subjective experience may not be the same as those which may make artificial life conscious. Indeed, even if it appears that some androids have all the characteristics of our brain that we attribute our conscious experience to, we could never rule out an unknown, further physical property of our brains, lacking in androids, which might make humans conscious but androids not.

A more answerable subset of this question concerns whether androids might experience sensations of suffering or happiness. In humans, the former is generally associated with behavior and situations which reduce an organism’s fitness to survive and reproduce in its environment, whereas the latter is associated with increased fitness. Maybe these properties can only arise in such circumstances, and by only manufacturing androids with strict limitations in the way that they process the state of themselves and their environment, we could ensure that they would be very unlikely to experience these particular sensations. This approach might at least help solve the ethical question of whether removing an android’s power source would be murder!

Greg Davis. Department of experimental psychology, Cambridge, England.

My approach to AI and physics of consciousness is much more serious than the question regarding ‘Will androids ever dream of electric sheep?’ As previous chairman of PhysComp 92 and PhysComp 94, it is clear that information (i.e. consistency laws) is a basis for all of physics, and my personal belief is that consciousness is a matter of information mechanics as well. Scientists know that virtual reality systems have very large computational demands, but lucid dreaming is the most precise thought-directed 'virtual reality' system that exists. Tapping the computational power of 'real intelligence' of humans is the challenge. We must not limit ourselves to current Western paradigm limitations, especially when Eastern knowledge of Chi appears to have many of the quantum-like computational properties that may open the door to the understanding of man’s consciousness.

Doug Matzke. The Seraph Foundation, Dallas, USA.

We seem not to be able to turn the brain off completely – it must be too hard to re-boot. Some large computer systems are like this; it’s a big deal to turn one off and on. So, I suppose a robot with a computer brain would be on all the time.

Why do we sleep and dream? Well, a practical reason is that with our poor night vision it would not have been wise in the early times to be blundering around in the dark. Better to lie down and be still till the sun comes back up. Robots wouldn't seem to have this reason for sleeping, though even for them there would be dead times.

But maybe dreaming serves a positive purpose beyond being the lava-lamp ticking-over of the resting brain. Maybe you're trying out conceptual groupings, playing with the data, doing what-ifs, defusing things by reliving them, running anticipatory scenarios, making great Jungian connections to the God within. Quite likely, dreaming is important. So maybe computers could be programmed, or could evolve to a program, which has them 'dream' for some of the time.

A chaotic screen-saver looks a bit like a computer dream perhaps, cf. My CAPOW, downloadable here.

And of course, if our computer-brained android robots are dreaming, why not dream of electric sheep? Real sheep too, of course, but maybe electric sheep a bit more than humans do, given the familial relationship between electric man and electric sheep. Do electric sheep have electric liver flukes?

I have never happened to dream of electric sheep myself, but maybe Phil Dick did. I dreamed, I think, of Phil Dick (non-electric).

"And when Phil Dick awoke, he wondered if perhaps he were an electric sheep dreaming he was Lao-Tzu."

Rudy Rucker. Mathematician and author of The Hacker and The Ants.

This question is a catchy version of the more mundane 'Will androids have emotional states like ours?' (We not only dream, we also feel pain, joy, etc.) And the answer, by my lights, is 'No' – at least not if the androids are mere computers 'hooked up' to the environment via sensors (eg; eyes) and effectors (eg; legs). This is so (as I've discussed in my book What Robots Can and Can't Be) because emotional states can’t be formalized as cimputational states. (What computer program can convey, to someone assimilating it, what it feels like to pet a sheep?) On the other hand, if by 'android' is meant a creature having a physiological substrate very much like ours (a substrate, I don’t think, is computational in nature), all bets are off – as they are, of course, in BladeRunner.

Selmer Bringsjord. Associate professor, department of philosophy, psychology, and cognitive science, department of computer science, Rensselaer Polytechnic Institute, New York, USA.

A definite yes, since my conscious computer chips will be fully conscious in a human, if not a superhuman, sense. Since mind is an integral part of the quantum properties of matter, quantum-based androids will be as human as we are. Indeed, they will dream. They may even have wet dreams!