"...our brains are at a very deep level quantum mechanical and non-deterministic and can never be accurately simulated by a deterministic machine in a reasonable time"
I'd argue against Penrose's assertion because there isn't really a need to accurately simulate the human brain in order to achieve machine thinking. Further, the human brain probably has a fail-safe mechanism "installed" that prevents achieving a state of complete
self-understanding in a reasonable amount of time without a major evolutionary event. It's about identifying, understanding, and copying patterns of thinking, not necessarily building computerized human brains.
Certainly, we are several years away from developing machines that can conceptualize and reason on their own (hopefully less than 500). For the short term, however, we
are getting closer to developing computer systems that can mimic some of the features of biologic systems (not necessarily sentient). Specifically, I'm talking about the tendency to maintain a constant state of equilibrium; Systems that are self-healing,
self-maintaining, able to produce positive feedback in response to negative perturbation, etc.
Computer systems with strong homeostatic tendencies are coming.