I am saying that it's severing the link between cognition and existence what happens in digital information processing. I agree that information processing also occurs in biological systems, thus perhaps making it reasonable to speak about analog information processing—but that is not relevant to my argument. I admit that my exposition is quite far from settling the issue. I'll try to provide further support here.

It is key to acknowledge the importance of time in true, enactive cognition (which entails sentience, consciousness and agency). But time is something alien to classical mathematics and logic; therefore, an artifact whose "intelligence" is based upon classical mathematics and logic cannot exhibit sentience, consciousness and agency.

Mathematically, time is a phenomenon that pertains to the noncommutative realm (you might want to look for Alain Connes work on von Neumann algebras). As far as we can tell, reality is noncommutative in this sense: quantum physics teaches us that the algebra of physical observables is noncommutative. This immediately leads to very deep questions regarding the nature of observers. I don't think sentience, consciousness and agency could exist in a timeless, classical (commutative) world. Thus, I don't think AI's will ever be, in the way that we are.