Steve Dufourny that this hard problem of consciousness has not still renormalised with concrete physics
the problem is not "hard" at all. however that requires first to understand what Consciousness actually is, at which point retrospectively - and very unfairly - it all becomes exceedingly blindingly obvious.
I fell into this trap. I dismissed almost everything I could find, learned the hard way, then came back and went, "oh. actually, all these people I dismissed earlier got about 75 to 85% right". a handful got as far as 95% right. and the last final pieces have taken me an entire year of continuous study to nail to the floor.
bottom line: it is a trap from which no reductionist or objectivist will ever escape. some examples:
the average of a sine wave is zero. therefore sine waves do not exist!!!!
a 2nd order partial differential equation must have 3rd order terms and above equal to zero (Jordan Order N=3). b b b but.. but I want a polynomial "best fit in a single variable!" (brain melts down at the unfairness of not having coefficients a b and c in one variable when the partial differential equation requires two)
these are the kind of easily understood absurdities which illustrate that the multi-dimensional case behind the phenomenon named "Consciousness" is easy to comprehend once you stop trying to look for an objective / linear / reductionist solution.
ya ain't gonna find one.
Mathematically, Consciousness "exists" at the edge of Critical instability Higher-order fluctuations. Reynolds numbers. Negentropy. the whole shebang.
Unfortunately there are well over 200 alternative uses and meanings for this one word. The Glasgow scale of consciousness is a medical term that is remarkably close to measuring the Negentropy currently "active" by a given (human) person, but a measure of Negentropy only tells you that Consciousness is present, not what it actually is.
To properly answer the question "can an AI ever be Conscious" the answer is "yes... if you, the programmer actually know what it is, enough to be able to actually implement a Conscious-compliant system!"
and as any Software Engineer knows, that requires a formal specification, a test suite, and a whole boat-load more all of which is straight out the window if your customer / client / manager / boss cannot explain to you clearly what it is that they want!
consequently we are rather amusingly getting a lot of money poured into efforts where the entire team have no idea whatsoever as to what Consciousness is! it's like cryptocurrency all over again. sigh.
I leave you with this to ponder, to ask if the Turing test - or any of the many variants of it - are ever going to make the grade.
https://www.researchgate.net/publication/390750336_Upgrading_the_Turing_Test_to_Consciousness