# Turing test tactics **URL:** \ **Category:** In My Humble Opinion\ **Created:** [March 4, 2011, 1:09pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475 "2011-03-04T13:09:38Z")\ **Posts on this page:** 15\ **Page:** 1
**Author:** ![willthekittensurvive](https://avatars.discourse-cdn.com/v4/letter/w/91b2a8/32.png) [@willthekittensurvive](https://boards.straightdope.com/u/willthekittensurvive)\ **Post date:** [March 4, 2011, 1:09pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/1 "2011-03-04T13:09:38Z")
What line of questions would you use to determine if an opponent was an AI or a meat bag? > **[Turing Test](https://xkcd.com/329/)** > > Hit Turing right in the test-ees. > **[A New CAPTCHA Approach](https://xkcd.com/233/)** > > They'd use that Futurama episode with Fry's dog, but even spambots cry at that. ---
**Author:** ![Double\_Foolscap](https://avatars.discourse-cdn.com/v4/letter/d/85e7bf/32.png) [@Double\_Foolscap](https://boards.straightdope.com/u/Double_Foolscap)\ **Post date:** [March 4, 2011, 2:07pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/2 "2011-03-04T14:07:30Z")
In real life (_Blade Runner_ style) or over instant messaging? In real life I’d ask regular questions but start behaving obnoxiously - picking my nose and flicking it in their direction, for example. Non-verbal clues would probably be a lot more telling in this hypothetical future I’ve just made up in my brain. Over IM, I’d ask what their favourite movie was and then start quoting it so see if they joined in. I’d also use bad spelling and slang, hep cat. ---
**Author:** ![Digital\_Stimulus](https://avatars.discourse-cdn.com/v4/letter/d/aeb1de/32.png) [@Digital\_Stimulus](https://boards.straightdope.com/u/Digital_Stimulus)\ **Post date:** [March 4, 2011, 5:30pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/3 "2011-03-04T17:30:36Z")
There are a few ways to go about it. Of course, the point is to highlight human capabilities that are difficult to program. A few that come to mind: [ul][li]Incorporate trivia and/or cultural references. Bonus difficulty if references are to foreign cultures or historical periods.[/li][li]Use wordplay in your responses, such as homonyms/homophones or other devices. Especially useful when used in conjunction with another item, such as the previously mentioned references.[/li][li]Another thing to do is to make heavy use of pronouns; it’s really hard to program subject resolution, particularly when dealing with natural language.[/li][li]Similarly, confuse the conversation with a variety of verb tenses or time frames.[/li][li]Probe multi-modal capabilities (colors, music, taste, etc.).[/li][li]Incorporate “back references” (that is, refer to topics or events that came up earlier) in the conversation. This forces the AI to do subject resolution in combination with maintaining a discussion memory.[/li][/ul] I suppose one could simply launch into a “_Who’s on first_” routine too. 😃 ---
**Author:** ![TravisFromOR](https://avatars.discourse-cdn.com/v4/letter/t/cc9497/32.png) [@TravisFromOR](https://boards.straightdope.com/u/TravisFromOR)\ **Post date:** [March 4, 2011, 7:03pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/4 "2011-03-04T19:03:25Z")
> [@Digital\_Stimulus](#): > > [ul] > [li]Another thing to do is to make heavy use of pronouns; it’s really hard to program subject resolution, particularly when dealing with natural language.[/li][/ul] Bill and Bob were working in a field, when he pushed him into a threshing machine. :dubious: ---
**Author:** ![42fish](https://avatars.discourse-cdn.com/v4/letter/4/b487fb/32.png) [@42fish](https://boards.straightdope.com/u/42fish)\ **Post date:** [March 4, 2011, 7:24pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/5 "2011-03-04T19:24:19Z")
> [@TravisFromOR](#): > > Bill and Bob were working in a field, when he pushed him into a threshing machine. > > :dubious: To be fair, he said he had it coming. ---
**Author:** ![Digital\_Stimulus](https://avatars.discourse-cdn.com/v4/letter/d/aeb1de/32.png) [@Digital\_Stimulus](https://boards.straightdope.com/u/Digital_Stimulus)\ **Post date:** [March 4, 2011, 8:06pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/6 "2011-03-04T20:06:01Z")
> [@TravisFromOR](#): > > Bill and Bob were working in a field, when he pushed him into a threshing machine. > > :dubious: Perhaps I’m being whooshed, but…are you really :dubious: about it? If so, you’re not putting enough thought into the point. Remember, the point is to confuse the thing; humans are really good at subject resolution. Besides the possibility of making a sentence inherently ambiguous (particularly if using the pronoun “_it_” with multiple referents), there’s also the matter of computational complexity (_cf_. [3SAT](http://en.wikipedia.org/wiki/Boolean_satisfiability_problem) as an NP-complete problem). The former would require the computer to ask for clarification, opening up many opportunities to confuse the program; the latter may, depending on complexity, result in noticeable time lag. ---
**Author:** ![Pleonast](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/pleonast/32/1183_2.png) [@Pleonast](https://boards.straightdope.com/u/Pleonast)\ **Post date:** [March 4, 2011, 8:52pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/7 "2011-03-04T20:52:50Z")
I thought this thread would be about passing a Turing test. Any good tips? 😉 ---
**Author:** ![engineer\_comp\_geek](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/engineer_comp_geek/32/504_2.png) [@engineer\_comp\_geek](https://boards.straightdope.com/u/engineer_comp_geek)\ **Post date:** [March 4, 2011, 9:46pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/8 "2011-03-04T21:46:13Z")
> [@Pleonast](#): > > I thought this thread would be about passing a Turing test. Any good tips? Here’s a tip. When Mr. Holden asks you about your mother, try not to shoot him. ---
**Author:** ![Ike\_Witt](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/ike_witt/32/283_2.png) [@Ike\_Witt](https://boards.straightdope.com/u/Ike_Witt)\ **Post date:** [March 4, 2011, 10:14pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/9 "2011-03-04T22:14:36Z")
> [@engineer\_comp\_geek](#): > > Here’s a tip. When Mr. Holden asks you about your mother, try not to shoot him. So, is Voight-Kampff a Turing test? ---
**Author:** ![BigT](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/bigt/32/12044_2.png) [@BigT](https://boards.straightdope.com/u/BigT)\ **Post date:** [March 5, 2011, 10:03am UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/10 "2011-03-05T10:03:34Z")
> [@Digital\_Stimulus](#): > > There are a few ways to go about it. Of course, the point is to highlight human capabilities that are difficult to program. A few that come to mind: > [ul][li]Incorporate trivia and/or cultural references. Bonus difficulty if references are to foreign cultures or historical periods.[/li][li]Use wordplay in your responses, such as homonyms/homophones or other devices. Especially useful when used in conjunction with another item, such as the previously mentioned references.[/li][li]Another thing to do is to make heavy use of pronouns; it’s really hard to program subject resolution, particularly when dealing with natural language.[/li][li]Similarly, confuse the conversation with a variety of verb tenses or time frames.[/li][li]Probe multi-modal capabilities (colors, music, taste, etc.).[/li][li]Incorporate “back references” (that is, refer to topics or events that came up earlier) in the conversation. This forces the AI to do subject resolution in combination with maintaining a discussion memory.[/li][/ul] > I suppose one could simply launch into a “_Who’s on first_” routine too. 😃 I’d also pay attention to what type of stuff could fool Watson on Jeopardy. You’d be surprised about how many things on that list that he was able to handle. ---
**Author:** ![madmonk28](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/madmonk28/32/20811_2.png) [@madmonk28](https://boards.straightdope.com/u/madmonk28)\ **Post date:** [March 5, 2011, 11:12am UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/11 "2011-03-05T11:12:14Z")
I’m cramming for mine tomorrow, anyone know the difference between a turtle and tortoise? ---
**Author:** ![MrDibble](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/mrdibble/32/114_2.png) [@MrDibble](https://boards.straightdope.com/u/MrDibble)\ **Post date:** [March 5, 2011, 12:49pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/12 "2011-03-05T12:49:30Z")
> [@Ike\_Witt](#): > > So, is Voight-Kampff a Turing test? Not really. I’m pretty sure the replicants \*were \*sentient by any Turing definition, they just weren’t empathetic in the same way humans are, and that’s what V-K tested. ---
**Author:** ![Digital\_Stimulus](https://avatars.discourse-cdn.com/v4/letter/d/aeb1de/32.png) [@Digital\_Stimulus](https://boards.straightdope.com/u/Digital_Stimulus)\ **Post date:** [March 5, 2011, 4:27pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/13 "2011-03-05T16:27:53Z")
> [@BigT](#): > > I’d also pay attention to what type of stuff could fool Watson on Jeopardy. You’d be surprised about how many things on that list that he was able to handle. Yeah, I know. Very impressive. Although I have to say that I didn’t pay too much attention to Watson beyond the fact that it trounced the humans (excepting Rush Holt, who deserves props for holding his own). Can you point me to a decent analysis of its play/weaknesses (preferably not regular news sources, but more technical)? ---
**Author:** ![Saint\_Cad](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/saint_cad/32/18907_2.png) [@Saint\_Cad](https://boards.straightdope.com/u/Saint_Cad)\ **Post date:** [March 5, 2011, 4:43pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/14 "2011-03-05T16:43:12Z")
But it it a true Turing Test if a human idiot can’t pass? Let’s say you and Subject J3 are discussing how llamas and alpacas are used for wool. You mention that llamas and camels are related and Subject J3 replies that “Camels have been around for years.”\* “Aha!” you think, “This must be a computer because a human would have said ‘thousands of’ or ‘millions of’ years. The computer obviously was obviously refering to when we talked about the recent fad of cashmere sweaters.” Great analysis, but you were really talking to Karl Pilkington. - Actual quote from the Jordan show. ---
**Author:** ![Digital\_Stimulus](https://avatars.discourse-cdn.com/v4/letter/d/aeb1de/32.png) [@Digital\_Stimulus](https://boards.straightdope.com/u/Digital_Stimulus)\ **Post date:** [March 5, 2011, 11:22pm UTC](https://boards.straightdope.com/t/turing-test-tactics/573475/15 "2011-03-05T23:22:07Z")
> [@Saint\_Cad](#): > > But it it a true Turing Test if a human idiot can’t pass? From [Stanford Encyclopedia of Philosophy](http://plato.stanford.edu/entries/turing-test/): > [@](#): > > The phrase “The Turing Test” is most properly used to refer to a proposal made by Turing (1950) as a way of dealing with the question whether machines can think. Important to note: the test is not to see if a (human) idiot can fail, it’s to see if a machine can pass. Very different things. Also note that: > [@](#): > > According to Turing, the question whether machines can think is itself “too meaningless” to deserve discussion (442). If you’re interested at all, you should really read Turing’s original paper, _[Computing Machinery and Intelligence](http://www.loebner.net/Prizef/TuringArticle.html)_; more than half of it is Turing raising and answering objections.