The Turing Test

1. Is the Turing Test a sufficient test? That is, if a machine passes the test, would you agree it is intelligent?

It depends on how we define intelligence. If intelligence is all about making judgments based on logical deduction, then passing the test might serve as an evidence of the machine being intelligent. However, human intelligence is supposed to be much more inclusive than this. For one thing at least, it should be able to cope with variations on its own when an encountered situation is not preprogrammed. An interesting example would be, on the “talk with God” website, when I typed in “Did you have breakfast?”, it replied as “no, I don’t have any breakfast, but I do have some gossip”. Apparently, the phrase “did you have something” is misunderstood here as “did you possess something”. While programmed perfectly for many other responses, when it comes to the transformation of a verb under certain context, the program could not generate appropriate responses for this variation. Although a machine or program similar to this might pass the Turing test satisfactorily, it does not meet the broader requirements of being intelligent, such as self-generated adaptation, etc.

Also, it seems to me that the analysis skills the test requires resemble what the many data analysis software can do nowadays. There is one major difference that some preprogrammed questions are asked. Other than that, the way it collects the responses, analyses them and provides conclusions is just similar to some statistics software. But we are not considering those software as intelligent.

2. Is the Turing Test a necessary test? That is, does a machine have to pass this test in order to be intelligent?

If a machine can pass such test, it is reasonable to believe that it has certain ability in information processing related tasks. But if the machine fails, it is also unfair to conclude that it is not intelligent, since even humans can make mistakes in tasks like this. In addition, I don’t see this test as a perfect simulation of the problems we come across in the real world. Therefore, this test need not be viewed as the criteria for intelligence.

3. Will a machine ever pass the Turing Test? Why or why not?

The test is not requiring highly generic information processing skills. So if the machine is equipped with programming that enables it to analyze the responses it encounters, passing the test should not be a hard thing.

4. Will a machine ever be intelligent? Why or why not? (This may or may not be the same answer to the previous question).

I’m convinced that machines will be intelligent eventually. While I can’t really think of any sci-fi like and inspirational ideas to support this argument, I would like to point out a very fundamental common character machine and human mind share, that is, despite all the other differences, they both run on electricity. The imitation of the passage of electrical signals should be successfully accomplished in the end.

Also, breakthroughs in at least two areas are necessary for this dream come true. One is the mechanism of how human brain is functioning. This area has been vexing scientists for ages, but humans should be able to know all about themselves finally. The other could be the appropriate medium or programming methods used by machines. It is highly possible that chips are not the most appropriate material that machines should be using when it tries to perform intelligently, or that the programming methods we are using right now has already limited some possibilities of machines’ being intelligent.