Let us return to our investigation of the theory of computationalism. I feel that our next topic of contention is best introduced by returning to naturalist Richard Carrier’s quote regarding the success of computationalism: “there is nothing a brain does that a mindless machine like a computer can’t do even today, in fact or in principle, except engage in the very process of recognizing itself.” Carrier’s assertion is well received. There are many things human beings can do that (seemingly) computers are also capable of (e.g., playing chess, carrying on a conversation etc.).
However, the question is immediately raised regarding whether or not computers actually carry out processes, or whether they in principle ever could, in the same manner a human brain does. Does a computer really “play” chess or does it just seem to? Does a computer carry on a conversation or does it just seem to? (Does anybody really believe they are having a conversation with Siri on their iphones?) The foremost defender of the view that a computer does not function similarly to human cognition is the naturalist philosopher John Searle. His thought experiment that attempts to demonstrate this assertion is known as the Chinese Room Argument.
For those unfamiliar with the Chinese Room Argument, it goes something like this. Imagine a unilingual English speaking man in a locked room. The room has a small “in” slot and “out” slot. Through this “in” slot is slipped Chinese symbols. The man in the room has a rulebook that tells him, in English, which Chinese symbols to send through the “out” slot based on the symbols received. The rulebook does not tell him the meaning of the symbols; it only tells him that if he receives a symbol that looks like “such and such”, he should respond with the different symbol “such and such”. To native Chinese speakers it would seem that the man inside the room speaks and understands Chinese. But, obviously, the man speaks no Chinese whatsoever.
So what exactly does the Chinese Room demonstrate about the nature of digital computers? Well, in the thought experiment the man is doing exactly what a computer does—though admittedly to a much more simplistic degree—namely, manipulating symbols in adherence with an algorithm in which only syntax is emphasized. But the man in the Chinese Room does not understand the semantics behind the symbols he is manipulating; similarly, if a computer is simply programmed to manipulate symbols based on pure syntax then how can a computer be said to understand the meaning behind said symbols? We have once again run into a problem of semantics—something that humans intrinsically possess and produce, but that computers do not. Edward Feser summarizes the consequences: running a program, of whatever level of complexity, cannot suffice for understanding or intelligence; for if it did suffice, then [the man in the Chinese Room] would, simply by virtue of “running” the Chinese language program, have understood the language.
So, once again, we see that digital computers constitute a difference in kind from the human brain. A computer runs syntactically defined programs and algorithms, yet does not develop any type of semantic understanding of the symbols it’s manipulating. In contrast, the human mind can run syntactically defined programs and algorithms, yet it maintains, throughout, a complete semantic comprehension—and much of this semantic comprehension is antecedent to any syntactical manipulation!
Let us return to Carrier’s comment above regarding the “abilities” of computers. While a computer can run programs that simulate and mimic human behavior, this does not mean that the ontology behind the computer is the same as a human brain. A computer can simulate the moves in a game of chess, but it is not “playing” in any relevant sense of the word. A computer can simulate a conversation with a person, but it is not really “conversing” in any relevant sense of the word. A computer doesn’t understand what a pawn does, nor does it understand what a person is saying to it. It is only programed to give a certain range (sometimes thousands) of syntactical outputs based on a certain range of syntactical inputs. This is not intelligence; it is a pure mechanical process of manipulation.
Not only do computers not think or reason, as we have seen, in any relevant sense, but they don’t really show any signs of intelligence at all. The whole endeavor of creating Artificial Intelligence is simply nothing more than a misnomer. University of Oxford philosopher Luciano Floridi states that, “we have no intelligence whatsoever to speak in terms of AI today as you would expect it from a cognitive science perspective.” I maintain that, in principle, we never will—at least not as long as the man in the Chinese Room is only a difference in degree from computers.
In the next post we will examine the question regarding whether or not computers exhibit intentionality—obviously one of the most important features of human cognition.