Because no matter how advanced of a computer hardware/software you build, you will not have a 'mind' in it. See https://en.wikipedia.org/wiki/Chinese_room

Do individual cells understand Chinese? What a flawed argument, it doesn't at all address the proposition that consciousness is an emergent property of computation. The guy would, stuck in his room for thousands of years, provide unknowingly the computational fabric for the a potential consciousness of the AI.

No, individual cells don't understand Chinese. Neither would individual transistors if strong AI was possible. But that's not what the argument is about. The argument refutes the notion that you could have a complex (in terms of both hardware and software) system that implements an actual mind. This complex system could act exactly has a real mind on the outside, but would have no semantic parsing (i.e. understanding) on the inside.