Searle questions Functionalism and the Turing test on the validity of it actually showing that there is artificial intelligence through the Chinese Room thought experiment. Searle shows that logical responses to inputs can be obtained without understanding any of what is going on other than you are to follow instructions. Searle’s argument is persuasive in that it takes into account what understanding truly is and passes the Turing test while proving that the test is not accurate.
Searle’s Chinese Room thought experiment is a persuasive argument for why the Turing test cannot be used to determine if artificial intelligence has actually been achieved. The Turing test and Functionalism requires responses that seem logical when conversing with an “intelligent” life form but never takes into consideration that those responses could be given in some manner to the being. The Chinese Room questions this by having an individual who does not know Chinese in a room where he is then handed cards with Chinese characters and then will have answer back using rules on how to formulate a proper response. The individual is clearly intelligent enough to follow the instructions carefully but not able to understand what he is doing or have any knowledge of what he is conversing about. This is a strong argument because a machine that is programmed works in the same manner, it follows the set of instructions created by the programmer and produces work that one could say only someone who is intelligent and can think for themselves would be able to produce.
The Chinese Room passes the Turing test in that it offers responses which seem to replicate thought and communication in a manner which would seem plausible for the individual in the machine to understand what was passing between the outside and them. By having the inputs and outputs in a different language represents the “language” difference between machine programming and how people interact. The subtle nuances of the rules being available to help translate showcase the modules of pre-programmed parts to complete certain tasks. The machine only truly understands the program and the programming code not the human languages that are displayed or fixed by the program. This makes it an extremely persuasive thought experiment since it questions the Turing test by using the slight nuances of computer programing to prove that artificial intelligence needs a better testing method. The program itself is able to run with a couple of possible directions, there is no true random response or any un-programed response. This is why the Turing test is passed via the responses but considered invalid due to the fact that the program is not actually understanding or reasoning the inputs and outputs.
In argument of Searle’s Chinese Room the Functionalists did say that there was no definite intelligence or lack of, but more like a grayscale with varying shades of intelligence.
“That this, we should not assume that either something has no mind at all, or it has a full-blown intellect and a complex emotional life just like normal adult humans. Minds might come in degrees, from relatively simple minds to much fancier minds like ours.” (Jacobsen, 82)
This shows that there is varying possibilities for intelligence; Searle appears to have an all or nothing mindset in his thought experiment when in fact it is potentially varying degrees of intelligence being shown. The mindset of all or nothing could be where there was miscommunication between Functionalists and Searle. While this weakens Searle’s argument about the accuracy of Functionalism it still does not weaken his argument about artificial intelligence. This is because strong artificial intelligence is supposed to be able to replicate the functioning of the human mind not just proving that it has some intelligence.
The Systems Reply seems to be playing around with definitions of understanding, from understanding the language itself with understanding the rules given. Understanding the rules does not make one aware of what they are conversing about or if they are even writing Chinese, all they know is that the program says to do this. Understanding should require that if the individual was asked about their response in their native language that they would be able to keep conversing, this individual would not be able to and therefore should not be considered knowledgeable in Chinese. The system is seen as a whole in this argument, but if the individual is doing all the work and still does not understand Chinese the system does not understand what it is doing.
The Chinese Room is important to the Functionalist Theory because it shows that the test that they had believed to be infallible was in fact fooled by a computer-like program. Functionalists believed that machines could pass the Turing test only if they actually had been able to achieve artificial intelligence. So to respond using a program of inputs and outputs that the individual in the machine could not understand showed that this was not intelligence so much as a cleverly designed program to appear as though there was some intelligence where there was none. Following the rules does not mean the individual understands the language and therefore there is no understanding in the system. Functionalists believe in artificial intelligence and for the purpose of the Turing test strong artificial intelligence, which means that they believe that machines would be able to think and have cognitive processing much like humans. So to have their test be proven wrong by a program-like thought experiment shows that they need to re-evaluate the Turing test and come up with something more persuasive than the Systems Reply.
In conclusion the Chinese Room thought experiment is extremely well thought out and persuasive because of the attention to detail and understanding is a factor of finding intelligence. It shows that passing the Turing test does not show reasoning or understanding and therefore is invalid.
Sources:
Jacobsen, Rockney. An Introduction to Theories of Knowledge & Reality.
Boston: Pearson Learning Solutions
image source: (figure 3)
taylorbyron 