5
$\begingroup$

Turing Test definition taken from wikipedia:

The Turing test is a test of a machine's ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human. In the original illustrative example, a human judge engages in natural language conversations with a human and a machine designed to generate performance indistinguishable from that of a human being. All participants are separated from one another. If the judge cannot reliably tell the machine from the human, the machine is said to have passed the test. The test does not check the ability to give the correct answer to questions; it checks how closely the answer resembles typical human answers. The conversation is limited to a text-only channel such as a computer keyboard and screen so that the result is not dependent on the machine's ability to render words into audio

But how is a Turing Test defined?

What type of questions are considered to be so good that trick the human judge?

What and who defines a human judge suitable to be a judge? (for example if the judge is 5 year old wouldn't he be more easily tricked than a 50 year old computer scientist?)

All those chatbots are considered to pass the Turing Test? Most people are not sure if the chat bot is an actual bot or not.

Improve this question
$\endgroup$
4
  • 5
    $\begingroup$ I recommend reading Turing's original article, where he defined his test: Computing Machinery and Intelligence. It's from 1950, but could as well be from now. $\endgroup$
    – john_leo
    Commented Oct 31, 2014 at 10:46
  • 1
    $\begingroup$ I'd recommend that @john_leo post this as an answer, because it is in fact the answer: the Turing Test is defined in that paper. All that aside, I suspect that what Turing had in mind was that a program that could fool any human judge, with the human judge allowed to ask any questions whatsoever, are fair game. Now, that doesn't mean that a wrong answer to "air-speed velocity of an unladen swallow" would mean that the program wasn't intelligent. $\endgroup$
    – outis nihil
    Commented Oct 31, 2014 at 20:32
  • 2
    $\begingroup$ @outisnihil Thanks. I'll gladly add an answer with Turing's original definitions later this day. $\endgroup$
    – john_leo
    Commented Nov 1, 2014 at 8:50
  • 1
    $\begingroup$ I should clarify what I meant by "any" human judge: the machine needs to fool whatever judge can be thought of. So in a way, a Turing test is a falsifiability test: the test only halts on failure. $\endgroup$
    – outis nihil
    Commented Nov 3, 2014 at 15:39

2 Answers 2

Reset to default
10
$\begingroup$

Although Alan Turing is of course a very important computer scientist, the Turing test is only superficially related to computer science. It is more related to philosophy.

As long as machines exist people have wondered whether it is possible to construct a machine that can think. But to answer this question we must first answer the question what it means -- what is thinking? Turing tried to circumvent this philosophical problem by changing the question a bit: the goal is no longer a thinking machine but a machine which is indistinguishable from a thinking being.

The Turing test was intended as a thought experiment to make the reader think about (artificial) intellingence in some way, but it appealed to a lot of people. So people started to organize Turing tests for real, and if you're doing real Turing tests, then of course you need real rules. Last year the Turing test organized by the University of Reading attracted a lot of attention because, according to the rules fixed for that particular Turing test, one computer program beat the judges. See here for the press release.

Improve this answer
$\endgroup$
1
$\begingroup$

The Turing test highlights that there is no scientifically rigorous or quantifiable definition/ measurement of "general" intelligence in Turing's time, over a half century ago, and despite large advances in AI and new theories/ understandings of intelligence, neuroscience, psychology, etc, (and some claims otherwise) that gap in scientific/ human understanding still exists today. ie there is still no universal consensus or definitive theory. so the test sidesteps that issue (that it cannot be defined objectively so far) and treats it as a subjective question e.g. similar to questions of law, where "judges" make decisions/ rulings based on law, history/ precedent etc, or e.g. "prize contests" like the Nobel Prize that have scientific committees/ judges.

The Turing test shows some connection/ similarity to the psychology theory of behaviorism that emphasized (roughly) treating organisms as black boxes with senses and actions, and not speculating so much on their internal "state". however it has long been downgraded in importance in psychology.

The worlds current leading/ foremost Turing test is known as the Loebner contest by its main sponsor and this is usually the results reported in the media of chat-bots fooling humans etc., it is held annually, has particular conventions and rules set up by a committee, and has modest cash prizes.

Improve this answer
$\endgroup$
4
  • 1
    $\begingroup$ This is a nice little piece about the Turing test but it does very, very little to answer the actual question. Please try to answer the question instead of just writing some stuff you know about a couple of the keywords included in it. $\endgroup$
    – David Richerby
    Commented Oct 31, 2014 at 17:15
  • $\begingroup$ DR why dont you be specific about which part of the question (which itself rambles somewhat) is specifically not answered (your comment is a nice little criticism of the answer, but does very, very little to actually/ tangibly/ constructively/ specifically improve on it or indicate how to do so, and is just writing some generic stuff about a couple of standard stackexchange complaints). also, the questioner has no complaints. $\endgroup$
    – vzn
    Commented Oct 31, 2014 at 21:07
  • 1
    $\begingroup$ You haven't actually answered even one single part of the question! There are five question marks after the yellow quoted text: your remarks address exactly zero of them. $\endgroup$
    – David Richerby
    Commented Oct 31, 2014 at 23:37
  • $\begingroup$ lol respectfully disagree :P $\endgroup$
    – vzn
    Commented Nov 1, 2014 at 1:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.