what is searle's main argument Machine functionalism?
Functionalism is the view that mental states are nothing but the mental functions being performed. That is, there is no underlying mental "substance," for the lack of a better term, other than the functions performed by a mind.
Therefore, bearing that in mind, one can argue that AIs can actually have mental power, since there is no underlying mental "substance," and consequently, if we were to, at a point in time, have a machine that can produce effects associated with mental states given the inputs associated with the same mental states, then it would be safe to argue that the machine has a mind, solely by virtue of the machine being able to function as a mind.
Searle, however, argues that functionalism as a theory of mind is incomplete. The actual Argument (analogical) is quite lenghty; however, I will summarize the crux:
A Mental state is nothing, but inputs and outputs, as posited by functionalist. Therefore, if one were to create a computer such that it takes in data(generally associated with mental state X) and produces outputs generally associated with mental state X, then that computer could be arguably said to to possess mental state X.
Searle, though, argues that we can surely say that the hypothetical machine is performing the functions associated with Mental state X; however, it is not entirely true that it sctually posesses Mental state X. For instance, when it is cold, Human beings shiver. Now, suppose we have a computer that shakes at temperatures generally considered to be shiver-inducing. Is it intuitive to say that the computer is feeling cold?
Searle argues, therefore, regardless of how many algorithms we install in a particular machine it will never possess Qualia, and consequently mind.
For searle, then, a mental state is not constructed merely on syntax but on semantics as well.
Citation: 'Is the Brain's Mind a Computer Program?', John R. Searle, Scientific American