I think Searle's Chinese Room argument is sophistry, for similar reasons to the ones you suggest—the proposition is that the SYSTEM understands Chinese, not any component of the system, and in the latter half of the argument the human is just a component of the system—but Searle does believe that quantum indeterminism is a requirement for consciousness, which I think is a valid response to the argument you've presented here.
If there's actual evidence that quantum determinism is a requirement, then that would have been a valid argument to make instead of the Chinese room one. If the premise is that "it ain't sentient if it ain't quantum", why even bother with such thought experiments?
But there's no such clear evidence, and the quantum hypothesis itself seems to be popular mainly among those who reluctantly accept materialism of consciousness, but are unwilling to fully accept the implications wrt their understanding of "freedom of will". That is, it is more of a religion in disguise.
I don't know anything about this domain but I wholeheartedly believe consciousness to be an emergent phenomena that arises in what we may as well call "spontaneously" out of other non-conscious phenomena.
If you apply this rule to machine learning, why can a neural network and it's model not have emergent properties and behavior too?
(Maybe you can't, I dunno, but my toddler-level analogy wants to look at this way)
