The premise is wrong. The basic physical science underlying AI needn't be like brain architecture at all, because AI is a different kind of information processing problem. That's the very nature of AI, and the reason why we rely on it. It's not an attempt to clone the human mind: it's an attempt to solve discrete information processing problems. If the goal is to clone a human mind it might be that we use the same architecture - but that's only true if there is only one architecture that can support a mind we recognize as human. Even that is a leap at present.
On an analytical level this premise assumes that there's only one way to design minds - but that's an absolutely radical assumption. It may well be true - but there's no evidence for it (and our own information processing theories suggest the opposite).
As for Integrated Information Theory, there are a lot of issues. The basic analytical problem is that it takes consciousness for granted. It doesn't help us explain why conscious exist because it's an underlying premise of the system that it exists. This is helpful for humans - we know we are conscious (without jumping into solipsism or radical scepticism). This is not helpful for machine intelligence at all. Anyway, to discuss this further would be to launch into a fun intellectual debate that may be outside this thread but I'm game for it.
IIT theory doesn't predict consciousness - it assumes it and tries to create a theory around its existence. A lot of it is based on entrenched assumptions about human neural architecture. But AI is about information processing - it's an entirely different way of approaching human cognition. Just assuming phenomenology of consciousness leads to total gibberish.
Addressing the inaccuracies in this post is beyond the scope of this thread, but I'd be happy to discuss in private. I will touch on a few things though:
1) yes, the theory is derived from introspection on what criteria are required to satisfy consciousness. That's why it is parsimonious. It assumes consciousness exists, yea, but rather it attempts to explain its existence in exactly the same way that evolutionary theory attempts to explain biodiversity through descent with modification. The theory is about the mechanism of consciousness, not WHY it exists in the first place. And honestly, the latter is a philisophical question more than a scientific one.
2) Like any good scientific theory, it provides a rigorous model which can make predictions that test the theory. Some of those predictions are quite counterintuitive, and specific to this theory and no other. And so far, not only has every major prediction panned out, but we have even USED it to predict patients that have locked in syndrome and those about to emerge from comas. Clearly, the theory is correct - although it may be incomplete. It has been criticized as a theory of protoconsciousness rather than consciousness, which is fine with me as it is still light years ahead of any other scientific theory on consciousness that exists.
3) The theory does make specific predictions about what would be required to create a true AI, yes - and it does so because the predictions it makes are some of the counterintuitive ones that I discussed. For example, it specifically predicts that you cannot create a true conscious AI with feed forward processing alone. No other theory on consciousness makes this prediction.
You somehow think that is a problem with the theory, but to the contrary - it is a benefit of it. It is falsifiable.
4) I don't know what you are talking about when you say "entrenched assumptions about human brain architecture". I am a neuroscientist, and I have found no such assumptions in the theory. Furthermore, there aren't that many assumptions anymore - the architecture has been extremely well studied and mapped out. We still have a ways to go, but this isn't new science, we have known about the canonical cortical and cerebellar circuits for decades dude.
I don't know what your background is, I'm guessing more philosophy than neuroscience. And that's fine, but IIT is not only a good scientific theory, it is the best one on consciousness that we have.