Minds, Brains, and Programs

Comment 1: http://zmhenkel-chi2010.blogspot.com/2011/01/chinese-room-blog.html
Comment 2:  http://gspotblogspotblogspotblogspotblogspot.blogspot.com/2011/01/comments-reference-information-minds.html

Minds, Brains, and Programs
John R. Searle
Taken for open peer commentary from Behavioral and Brain Sciences by Searle, John R.

The abstract lays out a few distinct points that Mr. Searle defends in the paper, namely, he wants to show that while there are computers that do have true understanding, what people consider Strong AI is not a computer program that truly understands as true understanding is only represented by "...the appropriately programmed computer literally has cognitive states and the the programs thereby explain human cognition." He attempts to do this in two ways, one by asking well known members of the AI community what their interpretation of strong AI is, and second by allowing others to make claims and then systematically and logically disproving them as fallacy. He then ends by talking about his views on understanding and reasoning and explaining in a systematic way, what he believes they really mean.

The first example he explains is that of the Chinese Room. It is an example involving a man inside a room with a set of instructions on how to reply to a set of Chinese symbols. The man is given instructions in English that he understands and this helps him to essentially learn that, given this set of symbols I am to reply with this set of symbols, still not knowing what any of the symbols actually mean. He demonstrates how this is nothing more than a simple computer program, if this then this, given this print this etc. While this would make it appear as though the man understood Chinese it is nothing more than being able to follow a program or a system of instructions on how to relate one set of input to a set of output, similar to a computer program. He then presents these arguments to many different groups and researchers of AI and they came up with many arguments as to how we might be able to create a machine that perfectly imitates Human behavior so much that we would believe it is reasoning. However despite the arguments, Mr. Searle breaks it down to a simpler example and shows how it is nothing more than following a system and not true understanding. He then takes arguments from various schools and publications and in the same manner shows how each of their claims is nothing more than an extention of the original Chinese Room example.

Mr. Searle then explains his views on what understanding and comprehension really mean. He even goes so far to say "Could a machine think? ...yes We (Humans) are precisely such machines" but he makes it clear that he does not want to confuse this with the idea of truly understanding or being able to rationalize. He talks about three points that show the breakdown between a computer program expressing an output to an input and a mind comprehending a situation. The first is that the closest thing that a program can do to understanding is memorization, which is not the same as understanding, as memorizing Chinese characters is not the same as being able to read Chinese. The second is that while a program has a formal set of states, the mind does not have a set of states that it must change in an out of, only if we were able to have a program that did not have a set of states for determining of behavior would we be able to call it reasoning. Third, that mental states are a product of the brain, whereas the program is not a product of the computer. Essentially, that reasoning must come from the hardware that surrounds it, and as a program is written by another it is not a true creation.

He closes by again talking about how he believes that while machines can think, it takes a special kind of machine. He goes on to talk about how true origins of thought come from biological processes in the body that create reactions to situations. However this is not to say that a computer program can be constructed to explain to us how this occurs, it is not just a computer program that is based on a series of input and output.

This is one of my most favorite topics to argue and I have gotten in many discussions with roommates and friends over what would constitute a Strong AI. Various TV shows and computers have tried to imitate this but in the end it is very hard to define something as being a Strong AI. I believe the closest example ( and now we are going to get really nerdy) is Data from Star Trek as he seems to have not only logical reasoning systems but also displays curiosity, rationalization, and even intimacy without seeming to have a pattern or anything controlling these states. The show uses reasoning that I am sure Dr. Searle would agree that makes it appear as though he is a strong AI but I would be very curious, because of his in depth understanding of the topic, if this is nothing more than an extremely elaborate Chinese Room scheme. To that point, I would also like to say that in the case of the Chinese Room, because a computer program essentially acts in the same manner, it is not possible for a computer program to think. If we want to call our brains a machine then it is clearly possible for a machine to think but I don't think any machine governed by a set of programs is able to do the same. I would like to think that the new computer 'Watson' that recently won a game of Jeopardy or 'Deep Blue' who was able to beat champions in a game of chess understand what they are doing, but as shown in the Chinese Room example they are nothing more than extremely elaborate programs that only respond to complex input and can formulate complex output as though they understand what they are doing.