Comments
Reference Information
Minds, brains, and programs. Behavioral and Brain Science
John R. Searle
Summary
A thought experiment concerning the concept of true understanding by a machine is presented in this paper. The paper contends that even if a machine were able to pass the Turing test, communicating with a native speaker appropriately, it could not be said to have a mind. The paper argues against the idea of strong AI, which would accept the machine as having a mind. Searle holds that a machine, using symbol manipulation as a grounding, cannot be said to truly understand or posses a mind.
Discussion
The paper sparks some interesting thoughts concerning the definition of a mind and the distinctions we make between human intelligence and artificial intelligence. The idea of a program being able to pass the Turing test would seem to indicate a “mind” to me. I would propose asking the machine a question of ethics or personality. To pass the Turing test it must have a response (an opinion), it must reason about itself and it must be convincingly human. Ask enough of these questions and the “mind” of the machine might emerge.
So would you say that you agree with Searle? While I do agree that it must reason about itself and be capable of having its own opinion to be considered capable of truly understanding, your last sentence leads me to believe that you think it could be proven to have a mind with further studies. I'm leaning more towards that it's impossible for computers to ever truly "think" on their own since they will always be programmed.
ReplyDeleteI think that even asking a machine a question of ethics or personality could be covered by a well designed Turing machine that does not really "think". It could simply like anything else have a programed response to your question, and not really understand what you were asking.
ReplyDeleteI think all possible questions and possible answers to those questions could theoretically be enumerated and provided to a machine, and that would still not prove that it can think; it merely follows instructions.
ReplyDeleteBut as far as the Chinese Room goes, is learning a language not symbol manipulation, even for humans?
The most basic class in Computer Science will teach us that programs are sets of instructions. As you said, computers are only able to follow a set of instructions, as well as carry out calculations, etc. I think it is clear that a computer cannot have a mind nor brain of its own, humans do, that's why they were able to build computers in the first place! ;)
ReplyDeleteThanks for all of the comments! I don't think I would ever consider a machine to have a mind, but I do believe we will be able to create an illusion which is very convincing. Combining that with programs which modify themselves, learn and evolve, I think we will one day have something which feels very much like a unique human being.
ReplyDeleteWhen it comes down to it, we are all sort of "programmed". We choose to do certain things given how we respond to certain stimuli. That being said, a machine does that as well. I think what sets us apart is that many times we act on emotion or irrationally. Sometimes we choose right, others we don't. Does a machine need to act irrationally (not based on a rule) at times to be considered human-like?
ReplyDelete@Chris Kam, Even if we create a machine that is able to act irrationally, it is still following a rule. It may not be a well-defined rule, but a rule none the less. Even if the computer makes a random choice, when given a question, that choice still has to be a preprogrammed rule.
ReplyDeleteI agree with you on the illusion of the concept of a mind when it comes to programming. Thus explains why we invest in making such things as AI, intelligence (learning) processes. It definitely makes technology more relatable and not so 'machine like.' :)
ReplyDeleteI think it is next to impossible to create a mind. A mind implies awareness and autonomous action, but computers are told what to do by someone else. AI lacks the ability of individual rationality.
ReplyDeleteI don't know if it is possible to fabricate a thinking machine. I think a better indication of thought is not to see if a machine can do what it was designed to do, but if it can improvise and come up with new solutions on its own.
ReplyDelete