4 ideas
A man who knows no Chinese sits in a room and matches incoming Chinese characters against a rulebook written in English, then passes back the output characters the rules specify. To people outside, his answers are indistinguishable from a native speaker's. Yet he understands nothing, which shows that producing the right outputs by following formal rules is not the same as understanding.
Weak AI holds that computers are powerful tools for studying and simulating the mind. Strong AI holds that an appropriately programmed computer literally has a mind and that its programs explain human cognition. Searle's argument targets only strong AI, conceding that simulation is useful while denying that simulation is duplication.
Objectors say the man doesn't understand but the whole system—man, rulebook, scratch paper—does. Searle replies that the man can memorize the rules and do all the calculation in his head, so he becomes the entire system. He still understands no Chinese, so there is nothing in the system that isn't in him.
Searle holds that minds are caused by brains, and that intentionality arises from specific causal powers of biological neural tissue rather than from any program it runs. A machine could think only if it had causal powers equivalent to the brain's, which formal program implementation cannot supply because the same program can run on anything. Understanding is a biological phenomenon like lactation or photosynthesis, not an abstract computational one.