I failed the Turing Test

I am at fourteen thousand feet and trying to read “The Emperor’s New Mind” by Roger Penrose. (Its cloudy at UKIRT). Perhaps it would have been wiser to try it at sea level at a more normal neural firing rate.

So early on there is all this stuff about the Turing Test. Apparently Turing’s original paper said that computers would pass 30% of Turing tests by the year 2000. Now I love my Macbook but it ain’t that good. So chalk that up with personal jet packs on the list of futurology failures.

You know the idea of course. A computer and a real person are hidden away in different rooms and a second real person asks both of them questions, by plain text only. If the interrogator can’t tell which one is the real person, then the computer has passed the Turing Test.

So it occurred to me that the weakness in this is that it depends on the hidden human. I definitely have some cousins who would fail the Turing Test. Or let a dumb computer slip through too easily. Anyhoo, mustn’t be arrogant. At fourteen thousand feet I think I will fail too.

Advertisements

15 Responses to I failed the Turing Test

  1. Alan Heavens says:

    This always struck me as a surprising proposal, as it is subjective, and so different from the rigour of Turing machines and the achievements at Bletchley Park. Perhaps it was deliberately challengeable to stimulate debate.

    Anyway, I enjoyed the post, even though I know perfectly well it was a computer that generated it.

  2. andyxl says:

    Tell me more about your enjoyment of the computer generated post.

  3. Martin E. says:

    My suggestion is not to bother with the 2nd half of Penrose’s book (Emporers new mind), it’s trying to explain free will with quantum mechanics, which just isn’t going anywhere; now the *first* half – an exposition of quantum mechanics – is terrific!

  4. andyxl says:

    If anybody wants to play Turing Test games, try iGod.

    I read about it on the Uncyclopedia

  5. If you think you can distinguish music written by a computer from human-made music, try Tom Collins’s web site: http://users.mct.open.ac.uk/tec69

  6. Alan Penny says:

    One of the reasons for developing chess-playing computers was to understand human thinking. But when a computer could beat the world champion, it wasn’t thought that humans were now understood, but that chess wasn’t the way to understand them.

    My guess is that when a computer can pass the Turing test, it will merely show that the test is not very interesting. Would you have any compunction about switching off a computer that passed the Turing test? Even though it begged you not to?

  7. andyxl says:

    Alan – have you seen the Spielberg film “AI” ? Of course you would switch off “a computer”, but would you switch off that Android, while it was staring at you with big doggy eyes ?

  8. andyxl says:

    Steve – I got around to checking out that Tom Collins computer-music website. Its amazing ! I assumed it was all going to be Stockhausen-y clanking, but in fact it sounds more like a combination of Bach and Keith Jarrett. Jeez. I can’t tell.

  9. […] My Turing Test post turned out more interesting than I expected. Playing iGod was fun, and just that; but […]

  10. Alan Penny says:

    Andy (May 6),

    I hope I wouldnt be appearanceist when deciding if there was a person in a machine.

    My point about the Turing test is that the first machine to pass the test will merely be a clever set of programmes very unlike what we think as a person, and so wont qualify as a person. Much as the chess machines can do chess but dont think like people.

    Of course this doesnt rule out there later being machines which we do think of people. And these would by their nature pass Turing. But since we dont now know what ‘being a person’ means – for example we dont even know what consciousness is – this seems a long way in the future.

  11. cellserf says:

    What do you think about watson the computer that will challenge humans at Jeopardy?

  12. andyxl says:

    You speak in riddles. Link please or other explanation ? ps Brit readers don’t know what “Jeopardy” is.

  13. Tony says:

    Programming a machine to play chess is interesting but has nothing to do with being human. A machine that can *learn* to play chess, picks up a few moves, plays and loses a few games then decides the whole damn game is boring and so goes off and watches Star Trek instead: now that would be an achievement.

  14. Alan Penny says:

    Cellserf (whoever he/she may be) is referring to IBM’s DeepQA (Question Answering) project to ‘take business intelligence to a new level’ (business intelligence – an oxymoron surely).

    This includes the ‘Watson’ computer for the Jeopardy! challenge.

    This is a good example of a ‘Turing test’ type machine that is not a person.

  15. […] intelligence, philosophy, robots, science fiction, turing test | No Comments  Andy has a post about the Turing test: how to determine between a software program and a human based on the answers […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: