The Turing Test and Proxies for Intelligence

In my language class, we have been talking about how to get at some abstract properties of language, such as the reading level of a document or the level of technical detail, and what shortcuts we can take to find these properties without needing the machine to fully understand natural language. These shortcuts are usually in the form of proxies, something which "speaks" for something else. For example, the amount of latin words in a document might be a good approximation of the level of technical detail in it. Certainly, it is not a perfect approximation (or it would be called a "calculation"), but it's a much cheaper and easier way of getting an answer, especially when compared to building a full language parser with knowledge base.

Then halfway through all this stuff, I realized something: that the Turing test is actually using conversation as a proxy for intelligence.

Isn't that interesting? Normally when I think of the Turing test, I think about the machine fooling humans, Eliza, sentence substitution, etc. Somehow in talking about Turing test I forget that it was supposed to be a test of intelligence for machines. Looking at the Turing test as an estimation of intelligence immediately raises an obvious question: is it a good estimation? How well does the Turing test actually represent intelligent behavior?

Then again, what is intelligence? As you might have guessed, the definition of intelligence has been in debate for a long time. For example, the following definition was signed by 52 intelligence workers:
a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings—"catching on", "making sense" of things, or "figuring out" what to do.
That definition, however, seems to say more about what intelligence is not than what it is.

Of course, the reason we have the Turing test in the first place is because intelligence is an extremely slippery creature to catch. No one can really define what intelligence is, and it's even harder to measure it. While the Turing test using conversations as a proxy, there are other proxies as well which are in use.

Example: the SATs. Even though now the letters don't stand for anything (according to Wikipedia), at one point it was supposed to test for the intelligence of its takers. The question is, how well does questions of analogy, math, reading, and writing serve as proxies for intelligence? The SATs do not seem to fit any of the capabilities ("among other things") as defined above, maybe except the ability to reason.

Another commonly mentioned proxy for intelligence is the Intelligence Quotient, or IQ. Again, simply because it claims to measure intelligence, it has a considerable amount of criticism against it. I have never taken an IQ test before, and so cannot sat anything about the fields it covers. It would seem to me, however, that any sort of test which can be administrated in any fashion would neglect some form of intelligence found in the real world.

I don't know if any of language, SATs, or IQ are good proxies for intelligence. I personally think that intelligence has to observing the environment, learning to adapt to it, and applying knowledge to affect the world. I don't think this is any more specific or scientifically testable than the definition above though.

I dare say I am not intelligent enough to think of a better definition.

1 comment :

  1. ...
    In Turing Test Two, two players A and B are again being questioned by a human interrogator C. Before A gave out his answer (labeled as aa) to a question, he would also be required to guess how the other player B will answer the same question and this guess is labeled as ab. Similarly B will give her answer (labeled as bb) and her guess of A's answer, ba. The answers aa and ba will be grouped together as group a and similarly bb and ab will be grouped together as group b. The interrogator will be given first the answers as two separate groups and with only the group label (a and b) and without the individual labels (aa, ab, ba and bb). If C cannot tell correctly which of the aa and ba is from player A and which is from player B, B will get a score of one. If C cannot tell which of the bb and ab is from player B and which is from player A, A will get a score of one. All answers (with the individual labels) are then made available to all parties (A, B and C) and then the game continues. At the end of the game, the player who scored more is considered had won the game and is more "intelligent".
    ...


    http://turing-test-two.com/ttt/TTT.pdf

    ReplyDelete