computer passes Turing test

[dragon]

Barely by 3% granted maybe they had a dumb judge still a remarkable breakthough

A programme that convinced humans that it was a 13-year-old boy has become the first computer ever to pass the Turing Test. The test — which requires that computers are indistinguishable from humans — is considered a landmark in the development of artificial intelligence, but academics have warned that the technology could be used for cybercrime.

Computing pioneer Alan Turing said that a computer could be understood to be thinking if it passed the test, which requires that a computer dupes 30 per cent of human interrogators in five-minute text conversations.
Read more: What exactly is the Turing test?

Eugene Goostman, a computer programme made by a team based in Russia, succeeded in a test conducted at the Royal Society in London. It convinced 33 per cent of the judges that it was human, said academics at the University of Reading, which organised the test.

It is thought to be the first computer to pass the iconic test. Though other programmes have claimed successes, those included set topics or questions in advance.

A version of the computer programme, which was created in 2001, is hosted online for anyone talk to. (“I feel about beating the turing test in quite convenient way. Nothing original,” said Goostman, when asked how he felt after his success.)

The computer programme claims to be a 13-year-old boy from Odessa in Ukraine.

"Our main idea was that he can claim that he knows anything, but his age also makes it perfectly reasonable that he doesn't know everything," said Vladimir Veselov, one of the creators of the programme. "We spent a lot of time developing a character with a believable personality."

The programme's success is likely to prompt some concerns about the future of computing, said Kevin Warwick, a visiting professor at the University of Reading and deputy vice-chancellor for research at Coventry University.

"In the field of Artificial Intelligence there is no more iconic and controversial milestone than the Turing Test, when a computer convinces a sufficient number of interrogators into believing that it is not a machine but rather is a human," he said. "Having a computer that can trick a human into thinking that someone, or even something, is a person we trust is a wake-up call to cybercrime.

"The Turing Test is a vital tool for combatting that threat. It is important to understand more fully how online, real-time communication of this type can influence an individual human in such a way that they are fooled into believing something is true... when in fact it is not."

The test, organised at the Royal Society on Saturday, featured five programmes in total. Judges included Robert Llewellyn, who played robot Kryten in Red Dwarf, and Lord Sharkey, who led the successful campaign for Alan Turing's posthumous pardon last year.

Alan Turing created the test in a 1950 paper, 'Computing Machinery and Intelligence'. In it, he said that because 'thinking' was difficult to define, what matters is whether a computer could imitate a real human being. It has since become a key part of the philosophy of artificial intelligence.

The success came on the 60th anniversary of Turing's death, on Saturday.

link

[Darth Tanner]

I'd have thought most chat bots would be able to pass as a 13 year old boy on the internet years ago.

5 minutes of conversation doesn't seem long enough to realise your speaking to canned responses... has the test always been to last 5 minutes?

[Channel72]

Meh... this really seems like it's cheating the "spirit" of the (rather subjective) Turing test.

A 13 year old boy, huh...

I guess I can write a Turing-test passing program that just emits random gibberish and claim it's actually a 2-year old human banging on a keyboard.

[Lagmonster]

TechDirt seems fairly certain this is bullshit:

Link here

[Ziggy Stardust]

I've never been a fan of the Turing test, anyway. It isn't a specific enough criterion to actually be empirically useful. Hell, unscrupulous scientists have tricked independent observers into thinking parrots (and other animals) had learned to speak English through clever and careful use of scripting. There's nothing remarkable about the ability to trick a random person into thinking they are talking to another person. It's a more or less worthless criterion; the only instance in which it could actually be useful is if you get 100 computer scientists/AI researchers who know some of the signs/exploits to look for trying to actively disprove that they are talking to a person.

[Starglider]

I've never been a fan of the Turing test, anyway. It isn't a specific enough criterion to actually be empirically useful... the only instance in which it could actually be useful is if you get 100 computer scientists/AI researchers who know some of the signs/exploits to look for trying to actively disprove that they are talking to a person.

It is actually quite effective when the judge is reasonably intelligent and analytical and there aren't any bullshit restrictions or staging. Five minutes (as originally specified) is a bit short but it's more than sufficient to do some easy-for-a-human but cognitively complex visualisation, common sense, creative and philosophical questions. 5 minutes each with a selection of 100 philosophers, programmers and psychologists is certainly the kind of protocol you'd want if you were making a serious academic claim of passing the test.

Alas the simplicity, fame and general popular appeal of the Turing test means that we get a lot of awful faked versions of it. That is instructive in its own way, i.e. as examples of the 'Eliza effect', but not in the originally intended sense.

[Ziggy Stardust]

It is actually quite effective when the judge is reasonably intelligent and analytical and there aren't any bullshit restrictions or staging. Five minutes (as originally specified) is a bit short but it's more than sufficient to do some easy-for-a-human but cognitively complex visualisation, common sense, creative and philosophical questions. 5 minutes each with a selection of 100 philosophers, programmers and psychologists is certainly the kind of protocol you'd want if you were making a serious academic claim of passing the test.

Fair enough. That is more or less what I meant to say with my last post, it was just kind of rushed because I'm idly checking the forums from an airport bar on a layover, heh. Just out of curiosity, would a "real" Turing test have more specific criteria (like a checklist of sorts) that they are looking for? I imagine they would, but I am having a hard time finding good info on the subject from my phone.

[biostem]

I skimmed the article, but didn't see any mention about how the conversations w/ the program were moderated; Was the judge able to talk about whatever they wanted and to switch topics or reverse the question-answer process mid-conversation?

I also think that the program claiming its age and lack of knowledge is a cop-out; By feigning ignorance, the judge's expectations are lowered.

[Tribble]

I imagine the success rate would be a lot higher if we judged it based on the way most people send text messages today. I Can Has Cheezburger?

[Terralthra]

Finally, computers have reached the point at which one time, with three average folks, a chatbot can accurately simulate a 13 year-old boy from Ukraine speaking English as a second language with a 33% pass rate.

Sigh.

[Starglider]

I imagine the success rate would be a lot higher if we judged it based on the way most people send text messages today. I Can Has Cheezburger?

Not necessarily, both understanding and generating plausible colloquial speech is harder than normal speech, because it is more context dependent (requires more inference to disambiguate and fill in implied details)

[Spoonist]

There are so many BS signals in this one that it definately falls into the scam/joke category of articles.
It's clearly badly machine-assist translated for instance.

has become the first computer ever to pass the Turing Test

nope it isn't, hence why the protocol has been changed a couple of times, Eliza is an internet meme for a reason, most NPCs in modern RPGs would pass the initial versions of the turing test (which fameously wasn't what Turing proposed by the way), as well as most corporate FAQ bots.

Alan Turing said that a computer could be understood to be thinking if it passed the test

nope he didn't, quite the opposite, he said it was likely to be impossible to agree to a definition of artificial 'thinking' so he did something completely different, which was an imagination game

which requires that a computer dupes 30 per cent of human interrogators in five-minute text conversations

That is just wrong

test conducted at the Royal Society in London

no they probably didn't, if so they would have published the result themselves...

said academics at the University of Reading, which organised the test

either its the royal society or it is the Uni of reading, pick one

It is thought to be the first computer to pass the iconic test.

Again, nope.

Though other programmes have claimed successes, those included set topics or questions in advance.

No they didn't

A version of the computer programme, which was created in 2001, is hosted online for anyone talk to.

In 2001, gimme a break...

Having a computer that can trick a human into thinking that someone, or even something, is a person we trust is a wake-up call to cybercrime.

quoted out of context for fake credibility yes Kevin most likely said those, but not regarding a limited turing of a 13yo character http://www.kevinwarwick.com/

Judges included Robert Llewellyn, who played robot Kryten in Red Dwarf

Of famed AI knowledge and ... ah forget it

[Spoonist]

lol found the source in Readings news
http://www.reading.ac.uk/news-and-event ... 82810.aspx

[Terralthra]

which requires that a computer dupes 30 per cent of human interrogators in five-minute text conversations

That is just wrong

You sure about that?

It will simplify matters for the reader if I explain first my own beliefs in the matter. Consider first the more accurate form of the question. I believe that in about fifty years' time it will be possible, to programme computers, with a storage capacity of about 109, to make them play the imitation game so well that an average interrogator will not have more than 70 per cent chance of making the right identification after five minutes of questioning.

[Eleas]

Buzzfeed weighs in. If Kryten is a reputable source, then so is Buzzfeed.

http://www.buzzfeed.com/kellyoakes/no-a ... uring-test

[AniThyng]

Transcripts from the public version of the chatbot.

http://www.scottaaronson.com/blog/?p=1858

Perhaps the tested version was jazzed up, but come on, this hardly seems any improvement over ELIZA.

[Borgholio]

That chatbot sucks. Is he actually proud of that? I've talked with spambots that do a better job.

[Lagmonster]

According to TechDirt's article, the guy who put this up has a long history of PR stunts and extraordinary claims about technological advancements.