Wouldn't a more "defined" question be whether or not we can accurately model the human brain(possibly to the molecular level) and make nearly 100% accurate predictions of all the decisions the subject would make? Basically, build an accurate predictive model of the brain. Like, say, predicting tomorrow's weather.

I'm not equipped to answer that, but there seem to be enough whizzes around here; is that problem theoretically unsolvable, or just very difficult? If it is unsolvable, then the only practical question is answered: The "real" brain is what it is and does what it does, and all the models are just approximations that give some answers, but never the whole picture. If, OTOH, is possible to build an accurate simulation, that would lead to some interesting questions...such as, for example, is it ethical to "scan" the brains of citizens in order to determine if they have strong criminal tendecies?