> How would you convince an intelligent being that could not feel pain that it exists at all?
Assuming the being is "reasonable" (in this context it would mean it's willing to accept that there exist concepts that it may not understand or directly experience, and is willing to trust us), we could just point out the chemical and electrical phenomena correlated with pain and say that it's something that causes a certain kind of feeling of discomfort. We would get in trouble if this being also can not feel general discomfort, but you're probably bound to hit a wall in understanding at some point anyway when communicating with an entity whose experiencing capabilities are wildly different from ours.
> The existence of pain falls outside the boundaries of scientific inquiry, I agree. But are you saying that it therefore doesn't exist?
Actually, my point about pain was that it does exist, precisely because it can be examined and explained within a scientific framework. If we couldn't do that, then we could say that for all practical purposes, as far as science is concerned, "pain doesn't exist".
The same is true for consciousness. What's difficult about it is that it's not a thing, there's no hormone for consciousness, there's no brain centre where it's localized, rather it's a process and a product both phylogenic and ontogenic so it's a lot harder to capture it and identify it, to put it "under the microscope". It's not some secret sauce to intelligence, it's a consequence of intelligence. And the most important part of the process is the dynamics of language acquisition (at least when we're speaking of conscious experience in Homo sapiens).
I could go into the details but I'm afraid my posts would explode in length. ATM I don't have time to dig for good online material on this, and I'm under the impression that the theory in time got derailed into developing some practical aspects concerning child cognitive development, verbal learning etc, and away from the hard meaty implications we're discussing here, so I'm reluctant to even attempt to go into that rabbit hole. But they're explicitly there (the books I mentioned discuss the issue at length). Interestingly, about 10 years ago I was doing some work on word-meaning and symbol grounding development and I was both glad and frustrated to see literature on computer modelling in this area full of operationally defined concepts from the theory but people were seemingly unaware that this work has already been treated in depth on the theoretical level because there were no references to it then, I'm not sure if anything has changed, I've since moved on to other things. For example, the Talking Heads model[1][2]. It's not about consciousness per se, and although the authors never reference the socio-cultural theory of cognitive development (a horrible name in this day and age, it tends to evoke associations to post-modern dribble, but nothing could be further from the truth), it can give you a good idea of some aspects of the dynamics explored in the theory because what is happening in the TH model is exactly what the S-C theory describes is happening externally during language acquisition (in broader strokes though).
As for the philosophical zombie problem, I'd like to retract what I said about it being nonsense. Actually, it's very useful in showing why worrying about subjective sensation of consciousness is completely useless in AI and is very much like asking how many angels can dance on a tip of a needle. On a very related note I'd add: people are severely underestimating the significance of the Turing test.
[1] http://staff.science.uva.nl/~gideon/Steels_Tutorial_PART2.pd...
[2] http://scholar.google.com/scholar?hl=en&q=talking+heads+expe...