The Vacuity of the "Turing Test"

The idea behind the Turing Test is that we must evaluate the possible intelligence of anything we encounter as though that thing were a black box.

So let us do a "Turing test" to decide who is the most knowledgable high school history student in the United States this year. Preliminary competitions have narrowed the field down two candidates: Jamal, who comes from a poor family, and Emily, who is from a wealthy one.

Per Turing, we must set each student within a black box from which that student's answers will emerge, and we are simply not allowed to inquire at all what is going on within that box. Jamal's parents, being poor, and trusting in the honesty of the contest, simply put Jamal in his box. But Emily's parents, knowing the way the world works, and being rich, hire a dozen top historians from around the world to sit in Emily's box with her.  Every time a history question is asked, Jamal answers the question himself, but Emily asks her team what the answer is, they tell her, and then her box "emits" that answer.

While Jamal does OK, "Emily" easily beats him in the contest. Per Turing, we must conclude that Emily knows more about history than does Jamal.

Similarly, per Turing, if we need to judge "Who knows most about when I must wake up," my mom or an alarm clock, we can't look at how either "system" came to wake me up: we can only say, "Well, given my mom came in my room at 6:30 and made a loud noise to wake me, and my alarm also made a loud noise at 6:30 to wake me, their knowledge is equal." And rabbit traps "know" the rabbit is there, since they trap it.

In short, the Turing Test is nonsense. It is not the way we decide if "meat machines" are intelligent, and it is no way to determine the intelligence of anything else, either.

11 comments:

  1. Pretty sure a crucial part of the TT (and regular school quizzes for the matter) is the subject's being cut off from external help...

    ReplyDelete
    Replies
    1. So, the computer can't rely on any programmers?!

      Delete
  2. "The idea behind the Turing Test is that we must evaluate the possible intelligence of anything we encounter as though that thing were a black box."

    This statement is incorrect. A Turing Test is a test of whether or not the responding entity is a program or an independent intelligence.

    ReplyDelete
    Replies
    1. Uh, yeah, and Turing says we must evaluate that responding entity as though the thing were a black box. So that statement is correct.

      Delete
  3. The aim of the Turing test according to Turing is to answer the question "Are there imaginable digital computers which would do well in the imitation game?"

    In theory a computer could pass the test while getting specific questions on history wrong.

    So when you say 'While Jamal does OK, "Emily" easily beats him in the contest. Per Turing, we must conclude that Emily knows more about history than does Jamal' then this makes very little sense in the context of the Turing test.

    If some Turing testers thought they were testing Jamal against a digital computer , but actually the other box contained Emily and a bunch of historians - then that probably would be a vacuous test !

    The paragraph about the alarm clock bears so little relationship to the real Turing test, its is impossible to categorize it as anything but totally irrelevant to alleged vacuousness of the test.

    ReplyDelete
    Replies
    1. 'The aim of the Turing test according to Turing is to answer the question "Are there imaginable digital computers which would do well in the imitation game?" '

      And then he thought we should deem them intelligent. Which is nonsense.

      Beyond that, you just demonstrate that again you have no clue what you are talking about! As per usual, rob!

      Delete
  4. This post demonstrates more clearly than I ever could that you do not understand the Turing test. You can't even get the mechanics right, as Silas notes.

    ReplyDelete
    Replies
    1. This comment demonstrates more clearly than I ever could that you DON'T KNOW WHAT AN ANALOGY IS. I could not possibly have "gotten the mechanics" of the Turing Test wrong, since nowhere have I attempted to describe those mechanics! I have offered some analogies, WHICH OF COURSE DIFFER FROM THE ACTUAL TEST, SINCE THEY ARE EFFING ANALOGIES, OH WEE-BRAINED ONE.

      Delete
    2. Oh, so now your argument is that some test not a Turing Test fails some goal not the goal of the Turing test.

      Delete
    3. Yes, you don't understand analogies.

      Delete
  5. Gene, thought you would enjoy this:
    "Of course if humans continue on our current trends Alan Turing may turn out to have been prescient after all. As human society approaches the Narcissism Singularity it may ultimately become impossible for a third party observer to distinguish between the nagging narcissism of circuits and the nagging narcissism of meat."

    https://zippycatholic.wordpress.com/2016/04/24/i-think-they-have-been-working-on-the-wrong-human-trait/

    ReplyDelete