Shena Hoffman
Zack Henkel
Reference information:
Title: MINDS, BRAINS, AND PROGRAMS
Author: John R. Searle
Published in Behavioral and Brain Sciences
Summary:
The Chinese Room thought experiment explores the notion of strong AI, the notion that computers can be said to have a mind and cognition. The experiment is as follows: an English-speaking person is locked in a room with a door. The person is given Chinese input through the door by someone outside, and by using instructions to manipulate the input, passes output in Chinese. The person inside does not understand Chinese, but the person outside is lead to believe that s/he is speaking to another Chinese-speaking person inside the room based on the output s/he receives, and thus the Chinese Room passes the Turing test. But Searle holds that a program cannot give a computer a mind or understanding and concludes that "programs are neither constitutive of nor sufficient for minds." "I can have any formal program you like, but I still understand nothing."
Discussion:
This kind of semantic discourse has always interested me. Initially I very much agreed with Searle, but after much thought, I am now inclined to disagree. that the Chinese Room itself, can define the very nature of learning a language. Consider that, to learn a language, people follow "instructions" for "input/outputs" until they are committed to memory. Here, "instructions" and "inputs/outputs" can include complete and comprehensive translations for the sake of argument. This is done such that the language becomes second-nature and can still understand Chinese even if s/he somehow forgot English, the original "instructions". Now, is the Chinese Room no different, assuming the instructions are also comprehensive? Given enough time (and discipline), the person in the room will begin to recognize the appropriate outputs associated with the given inputs and will eventually be able to provide the correct or desired outputs without use of the original instructions. As such, it is not true that the person "understands nothing", as Searle stated.
On the other hand, I agree that computers and programs do not inherently have minds, regardless of how they behave or appear to behave. This is because I cannot confidently say that computers possess consciousness, sapience, self-awareness or even emotions, all of which I consider essential aspects of a mind, in addition to intelligence.
Regarding the first paragraph of your discussion, I think you make a valid point here. My counter-argument would be this though: After the person has been "given enough time", they have processed the instructions for so long that they no longer need the instructions. In this case, you are saying that it is not true that they "understand nothing", but the person still does not know how a Chinese symbol relates to a real-world entity. I.E. when shown the symbol 三明治, the man in the room does not it says "Sandwich". So what DOES he understand then? Just something to think about. I like your critical thinking on the subject though.
ReplyDeleteI would have to disagree with the notion you presented in your discussion where you claim that "given enough time," a computer can truly begin to understand what it is doing. For example, consider the question "How are you?" While the computer may know a couple of phrases as to how to respond this to this inquiry, does the computer truly understand that this question is about its conscious state of being? And if computers don't possess consciousness, how can it respond to such an inquiry other than randomly picking a phrase out of a bag?
ReplyDelete