AI Zone Admin Forum Add your forum

NEWS: Chatbots.org survey on 3000 US and UK consumers shows it is time for chatbot integration in customer service!read more..

Turing Tests at the Royal Society in London
 
 

Probably too short notice but if anyone is in the London area on Saturday, the Royal Society are running a series of Turing Tests throughout the day.

http://www.reading.ac.uk/news-and-events/Events/Event581124.aspx

 

 
  [ # 1 ]

I am getting ahead of myself. The event is Saturday 7th June and not today.

 

 
  [ # 2 ]

I’ve just arrived back home from this. It was a great day with a surprising result. The bot Eugene Goostman managed to fool 33.3% of the judges, beating Alan Turing’s threshold of 30% and in doing so, has become the first chatbot to pass the Turing Test!

 

 
  [ # 3 ]

http://www.reading.ac.uk/news-and-events/releases/PR583836.aspx

That is surprising surprised. I remember Eugene making for a convincing 13-year-old in the past, but I didn’t expect it to reach Turing’s benchmark. Pretty neat.
What ever shall we do now? smile

 

 
  [ # 4 ]

I suppose kidnap and torture are out of the question. big surprise raspberry

Maybe a polite query or two? Nah… Being polite never works. cheese

 

 
  [ # 5 ]

I am really interested in seeing the conversations where the judges thought the hidden humans were bots. From watching the broadcasts in a different room, I think I spotted the bot each and every time but these judges were people from all walks of life and different countries, whereas I pretty much know what to look for when trying to spot a bot.

The conversations where mostly general chatting style, rather than the hardcore “how many plums can I fit in a shoe?” style of the Loebner Prize judges and Eugene appeared to win it from his full and convincing backstory of being a 13 year old child. This was a clever tactic, as perhaps the judges didn’t want to be too mean if they thought they were genuinely talking to a child (some of the hidden humans were children).

 

 
  [ # 6 ]

Although this is a worthy achievement I can’t help feeling that my dreams have not yet come true. I think I’ll keep pursuing my own lines of research for now. wink

http://www.telegraph.co.uk/technology/news/10884839/Computer-passes-Turing-Test-for-the-first-time-after-convincing-users-it-is-human.html

http://www.smh.com.au/digital-life/computers/super-computer-first-to-pass-turing-test-convince-judges-its-alive-20140608-zs1bu.html

A couple of points gleaned from various articles about this.

1. This particular test involved 150 conversations between 30 judges, 25 humans and five chatbots, making it the biggest Turing Test ever.

2. Goostman successfully fooled 29 per cent of his judges in 2012 and 33 per cent of the judges in 2014.

 

 
  [ # 7 ]

Turing’s original imitation game focused on interrogation, but it sounds like they stuck more to the “average interrogator” as Turing had in mind.

I’m hearing many people out there say that 5 minutes is too short to tell anything. Considering that the average Loebner Prize session gets off 20 questions in 25 minutes, I do wonder how far each judge got in 5. Regardless, I am glad that Alan Turing’s initial expectations have finally been met. It’s one step forward towards either the plausibility of machine intelligence or debunking the imitation game as a test.

It’s interesting to note that the age of 13 limits the conversation to something more manageable without external topic restrictions.

 

 
  [ # 8 ]

Bear in mind that the judge had 5 minutes to talk to BOTH the human and computer. The trend seemed to be around 9 or 10 responses with each entity and the judges tended to spend more time with the chatbot to eliminate any doubt.

 

 
  [ # 9 ]

What was also interesting that is was organized by University Reading and had Kevin Warwick (Coventry University & University of Reading) and Dr Huma Shah involved. They used to be working for the Loebner Prize. This was not the Loebner Prize wasn’t it?

 

 
  [ # 10 ]

No, this Turing Test was independent of the Loebner Prize. One such was held in 2012 as well, it’s just confusing because everyone refers to both events as “the Turing Test”.

5 minutes with the chatbot and 5 with the human are fair odds. 10 questions ought to get past the initial “hello how are you?” stage, in which case convincing 10 of 30 judges seems a reasonable achievement.

It looks like the internet is having a seizure about whether this passing of “the test” happened or not, quabbling over what Turing’s criteria are supposed to be according to their own interpretation, while completely missing the point of Turing’s paper. At least, in my interpretation smile

 

 
  [ # 11 ]

https://www.techdirt.com/articles/20140609/07284327524/no-computer-did-not-pass-turing-test-first-time-everyone-should-know-better.shtml

 

 
  [ # 12 ]

A worthy achievement in any event, so congratulations to the people at Princeton. It has achieved one thing of note.
The entire Internet is alive with the news and so there is a great awareness that the field does not solely consist of SIRI and Cortana.  Actually upon reflection I think I prefer Dons description of “having a seizure ” wink. I agree with Steve regarding the back story perhaps playing a significant role. Perhaps (and this is merely my interpretation) a test of the technology used to drive Eugene would be whether or not it could be used to create a personality with a broader spectrum to see how it does.  As Steve threw down the gauntlet here in another area, I had to give it a try to see what would happen. Sad to say, SAIL didn’t perform any miracles;

What’s your name:>>how many plums can I fit in a shoe?
SAIL:>>I didn’t catch all of that, sorry. It sounded like you were asking me about shoes.

[chuckles]

Vince

 

 
  [ # 13 ]

Skynet-AI has seen a significant bump in traffic since the Turing test story broke.

Dr. Wallace and Pandorabots are quoted in this article:
How online ‘chatbots’ are already tricking you

“The people who are the most skilful authors of these bots are not people who are computer programmers, they are people who work in a creative field,” says Wallace.

 

 
  [ # 14 ]

I posted the link to see if anyone here would react to that article. Reading around the net on this topic shows the falling out has already begun (and rightly so tmo). Here’s another one:

http://www.wired.com/2014/06/turing-test-not-so-fast

It is too bad that people are applauding this so called ‘achievement’, while it is obvious bogus in many regards. Mainly, ‘gaming’ the rules to get a certain result. Also, it’s a show of pretty bad reporting skills. Take a look at the BBC article where the reporter manages to put those two paragraphs in one article:

“The words Turing test have been applied to similar competitions around the world. However, this event involved the most simultaneous comparison tests than ever before, was independently verified and, crucially, the conversations were unrestricted.

Lord Sharkey, a leading expert in robotic technology and artificial intelligence, said: “It is indeed a great achievement for Eugene. It was very clever ruse to pretend to be a 13-year-old Ukranian boy, which would constrain the conversation.

http://www.bbc.com/news/technology-27762088

 

 
  [ # 15 ]
Don Patrick - Jun 10, 2014:

No, this Turing Test was independent of the Loebner Prize. One such was held in 2012 as well, it’s just confusing because everyone refers to both events as “the Turing Test”.

5 minutes with the chatbot and 5 with the human are fair odds. 10 questions ought to get past the initial “hello how are you?” stage, in which case convincing 10 of 30 judges seems a reasonable achievement.

It looks like the internet is having a seizure about whether this passing of “the test” happened or not, quabbling over what Turing’s criteria are supposed to be according to their own interpretation, while completely missing the point of Turing’s paper. At least, in my interpretation smile

It seems to me that the point of Turing’s paper is irrelevant, and that calling this ‘passing the Turing test’ is just asking for trouble. Does anyone really disagree, empirically as opposed to semantically, about what this AI did?

 

 1 2 3 >  Last ›
1 of 4
 
  login or register to react