1. Is the Turing Test a sufficient test? That is, if a machine passes the test, would you agree it is intelligent?
In my opinion, the Turing Test is not a sufficient test, which means a machine is not necessarily intelligent even if it passes this test. The Turing test has several inborn defections. First of all, the standards or guidelines for asking questions to both the machine and the human interviewee are not clearly specified in the test. As a result, the questions asked are likely to be inscrutable to both parties or too easy/difficult to either of them. In this sense the fairness of the Turing Test becomes questionable. Second, the intellectual level of the two humans participating in the test, the interrogator and the interviewee in the test, is neglected. I believe this factor also largely influences the test results. In addition, this test merely focuses on the test results without displaying the genuine thinking of process of the participants. To this end, using Turing Test to prove a machine to be intelligent is unpersuasive.
I believe Searle’s “Chinese Room” stands as an excellent counter example to the Turing Test. According to such model, as long as we input a computer program of operating rules and procedures into a machine, it can complete some seemingly intelligent acts. Searle argued that programs like this are entirely syntactical but minds have semantics. As syntax differs from and by itself insufficient for semantics, programs are therefore not minds (Searle, 1997). Searle also noted that the computer program lacks intentionality, consciousness and other mental phenomena, which are caused by the physical-chemical properties of actual human brains (Pinker, 1997). Therefore I think even if a machine passes the Turing Test, it does not mean it is indeed intelligent, the test results only indicate that the machine successfully follows the appropriate programs input into it by its intelligent human creators.
2. Is the Turing Test a necessary test? That is, does a machine have to pass this test in order to be intelligent?
As far as I am concerned, the Turing Test is not a necessary test because a machine does not have intelligence. From the perspective of performance level, a machine is capable of completing more and more intelligent tasks that originally can only be done by human beings: playing chess, writing novels and melodies, proving a mathematical theorem, facilitating diagnosis, recognizing human voice and so on. However it is unarguable that the performance level of a machine is not on a par with that of human beings at all.
According to Pinker (1997), intelligence has certain characteristics: to make decisions rationally by sets of rules; to pursue something in the face of obstacles and to use rational rules to attain the goal in a variety of ways. A machine cannot make decisions or even think by itself. It just appears to make decisions according to the programs input into it by humans. It cannot keep pursuing the goal when a program fails, either. Nor can a machine discover or invent different ways to achieve the goal as human intelligence does. Furthermore, I believe a machine does not have emotions, which are considered as manifestations of intelligence. Accordingly I believe the Turing Test is unnecessary for judging the intelligence of a machine.
3. Will a machine ever pass the Turing Test? Why or why not?
Personally I think there is no way for a machine to pass the Turing Test. As mentioned in the first question, there are a number of the inborn drawbacks of the Turing Test. Consequently there would probably be very tricky questions for a machine. The first type of tricky questions is the illogical ones. For example, a question like “My mom tells me that she saw a giraffe is flying in the sky, how do you think about it?” makes no sense at all to a machine. Questions with cultural connotations must be difficult to a machine, too. How can a machine reacts to the question like “What do you think of the ‘super girl’ phenomenon in China?”(“Super Girl Singing Contest” is a similar TV show to American Idol and it is very popular among Chinese audiences). In addition, questions including puns are tricky for a machine, too, say, “Time flies like an arrow, fruit flies like a banana. How do you think about it?”
Millions of programs can be input into a machine to facilitate answering questions in the Turing Test, unfortunately, they would never be enough to tackle all the problems generated by human intelligence.
4. Will a machine ever be intelligent? Why or why not?
I am a pessimist that I think a machine can never be intelligent. Webster dictionary defines “intelligence” as “the faculty of acquiring and applying knowledge”. Human beings learn by trial and error. They are capable of constructing their knowledge by interacting with each other and with the environment. A machine, on the other hand, can take in the entire content of Encyclopedia Britannia with its huge memory, however it obviously lacks both the intention and ability to actively learn or apply knowledge to deal with the problems in contexts.
Moreover, one of the major formats of representation used by human brain is mentalese, “the language of thought in which our conceptual knowledge is couched” (Pinker, 1997: 90). I think a machine lacks this medium that it is unable to capture content or gist. This makes the abstract thinking and innovation impossible for a machine as compared to a human mind.
In addition, a machine does not have emotions, which are critical for the forming of intelligence. Emotions drive human being to learn, so if a machine is implanted with an artificial mind, it may begin to learn by communicating with the environment instead of copying the human brain passively. Unfortunately as far as I know, emotions are one of the most difficult research areas in the world. As a result, the research question like “how to transform emotions into a program that can be input into a machine?” becomes even trickier.
Reference
Pinker, S. (1997) How the Mind Works, W.W. Norton and Company: New York.
Searle, J. R. (1997) The Mystery of Consciousness, New York Review Books: New York.
No comments:
Post a Comment