Views into the Chinese Room: New Essays on Searle and Artificial Intelligence [Paperback]

The Chinese Room Argument (CRA) has nothing to do with the speed of computers or any future developments in artifical intelligence (at least as understood as following from Turing). The CRA is a purely formal argument intended to refute the claim that computers (defined as Turing machines) can think, or can understand, or are minds solely by virtue of their formal description. (This claim is the essence of "computationalism," after Turing's original formulation.) The CRA is that: 1) Syntax is not semantics. 2) The implemented synatactical or formal program of a computer is not sufficient to generate semantics. 3) Minds have semantics. 4) Therefore, computers (so defined) are not minds/cannot think/do not understand because they are not sufficient to generate semantics.

For example, the concepts we employ to think and the words we use to speak have meanings. But there is nothing in computationalism as syntax that has any meaning whatsoever. Whatever meaning an implemented formal program has results from its being programmed or interpreted by us. Syntax (e.g., a computer program) has no causal powers. Whatever causal powers computers have (e.g., to fly airplanes) results from our programming and our assigning interpretations to the electrical charge insides a chip, not from the program in itself.

The chapters in Views Into the Chinese Room attack different aspects of the CRA. But they address it as an argument that stands or falls on the truth of the premises and the validity of the inference, not on engineering questions such as the speed of computers, which are irrelevant. Searle believes that there are, in fact, thinking machines -- we human beings are biological machines that think. And he believes that there also could be artificially made machines that think. The CRA is meant to show only that an implemented computer program by itself cannot generate mental content or semantic content.

For a clear explanation of the CRA, see chapter 15 of this book, by Stevan Harnad, the editor of The Behavioral and Brain Sciences, where Searle's original paper appeared twenty years ago. Do not rely on reviewers who do not understand the argument in the first place.

This book isn't as much about 'how computers think' as it is about how computer scientists think - and feel.

On page 52 John Searle states that CRA (Chinese Room Argument) rests on two "fundamental logical truths":

1.) Syntax is not semantics (An English speaker could learn to recite passages from the Koran in Arabic yet have no understanding of their meaning.)

2.) Simulation is not duplication (A computer program could simulate digestion but that doesn't mean it could digest a pizza and beer.)

That's it! It just doesn't seem like rocket science to me. And I've never seen such a battle of straw men. Straw man 1: the Turing test proves human consciousness; straw man 2: CRA proves that machines can't think; straw man 3: computation is not part of our cognitive process - and so on. Professor Searle said in a lecture that it took him five minutes to develop this little thought experiment. So what's all the fuss about? I think I found the answer on page 295:

"One of the charges that had been laid against Searle by his critics had been that his wrong-headed critique had squelched funding for Artificial Intelligence..." - Stevan Harnad (p295)

I have no idea what goes on in academia with research funding and all. But I have a sneaking suspicion that this is the source of much of the hyperbole about `thinking machines'.

Terry Winograd, computer scientist at Stanford, uses the postmodern technique to counter Searle's argument. The meaning of the word `understand', he claims, is a social construction. Therefore it is erroneous to believe that there is a right or wrong answer to the question: does the computer `understand' the language it is using?

"The error is not one of flawed logic in the argument about artificial intelligence, but is more fundamental and more pervasive. It has to do with the basic orientation we take towards the truth or falsity of statements in natural language." - Terry Winograd (p80)

On page 94 one of the editors of this book, John Preston, questions Winograd:

JP: "You say that all uses of the term `understand' and its cognates depend on certain concerns and perspectives. Isn't there a problem of self-reflexion [sic] here? Mustn't it be that your own uses of those terms are right only for certain concerns and from certain perspectives, but wrong for and from others?"

TW: "You correctly point out that if I am to be consistent in being relativistic, then I must apply it to my own statements as well as to the ones I am analyzing. Indeed, my own evaluations of attitudes are matters of social construction that is not the same as individual subjective opinion, but is grounded in the social discourse rather than in an appeal to objective truth." - Terry Winograd (p94)

What I see here is a person operating at the very, very highest level of academia resorting to the typical California postmodern response, `It's all relative man.' It's comical. Like a father saying, `Don't you try to tell me my baby doesn't understand English.'

Nevertheless, this book explores one of the most fascinating philosophical questions of our time: will we someday be able to build machines like ourselves that will replicate human consciousness? Considering the enormity of that question, it is no wonder the overwhelming response to Searle's CRA has sometimes taken on the character of religious passion.

The human brain evolved to assist the survival of its owner while the owner navigated the dangerous jungles and forests of ancient times. Its ability to extract patterns from the information provided by the retina and optic nerve is quite phenomenal. The process by which your brain is recognizing my words and understanding my meaning is astounding.

Yet if you are asked to act like a computer by reading numbers, moving paper tape, erasing things and following instructions given on the paper tape, you will prove to be one of the slowest computers in the world. The original word `computer' referred to a man sitting in a room with paper, pencil and eraser. These human `computers' were replaced by machines a long time ago because they are too slow.

In summary, humans are fast and intelligent at being humans but slow at being computers. In the Chinese Room Argument, John Searle states that although we have a human mind which could otherwise be used to understand Chinese, this particular human mind does not in fact understand it. Given this stipulation, the human mind's ability to process language cannot be used and the only method of "understanding Chinese" is left to the "Chinese room" which consists of a computer run by the very slowest of CPUs, the human being sans abacus, sans calculator, sans silicon chips and sans hope.

The Chinese Room Argument is a trick argument that proves nothing. The computer room is so slow that it cannot ever think or understand Chinese. On the other hand, this doesn't say anything about whether a high-speed computer with the memory and processing power of the human brain might one day speak and understand Chinese quite well.