15000

User avatar
Wock
Posts: 2540
Joined: Fri Aug 24, 2007 8:09 pm
Location: Gallatin, Tennessee (USA)
Contact:

Thu Jun 12, 2014 6:02 pm Post

pigfender wrote:
Jaysen wrote: she is a <redacted> year old 9th generation American. Same is true of my mother

You married a woman the same age as your mother?


No. I think you misread that. He is from the South. He married his mother.
The wheel is turning but the hamster is still dead.

User avatar
Wock
Posts: 2540
Joined: Fri Aug 24, 2007 8:09 pm
Location: Gallatin, Tennessee (USA)
Contact:

Thu Jun 12, 2014 6:05 pm Post

Jaysen wrote:
nom wrote:I'm not impressed with a Turing test that involves a computer pretending to be a non-English speaking child with poor keyboard skills interacting in broken English via text
so basically the fact that a "learning system" was able to learn enough to simulate 75% of the internet population isn't passing the turing test...

I guess the question of passing is predicated on what our definition of human interaction in a given medium is expected to be. If the test was state as "convince us that these txts come from a human", then my wife would fail. Heck, 3/4 of the time I have to call her to ask her what her last txt was. And she is a <redacted> year old 9th generation American. Same is true of my mother (using an iPhone) and in laws (on droids).

My point is that academics forget about the real world too often. I think the test committee may have actually applied the following question to determine the "pass" for this test:
supposition wrote:Would this fool a normal person?

If they answered "yes" then it passed the test legitimately in my opinion.

Which means we are doomed on many different levels...


Here is a video document of the new turing test

http://www.trutv.com/shows/the-carbonar ... laces.html
The wheel is turning but the hamster is still dead.

User avatar
pigfender
Posts: 2912
Joined: Tue Oct 12, 2010 10:25 am
Platform: Mac, Win + iOS
Location: I share a head with a great many personalities
Contact:

Fri Jun 13, 2014 9:48 am Post

Jaysen wrote:Only you and vic-k would have gone in that direction...


The REAL reason Vic didn't write a chapter in the NiaDs!
*gasp*
http://www.pigfender.com | http://www.novelinaday.com
"Some dice only have sixes." nom, 19 Oct 2013
Image Image

User avatar
nom
Posts: 1927
Joined: Sun Aug 31, 2008 12:02 am
Platform: Mac + iOS
Location: Melbourne, Australia
Contact:

Fri Jun 13, 2014 1:02 pm Post

Jaysen wrote:so basically the fact that a "learning system" was able to learn enough to simulate 75% of the internet population isn't passing the turing test...

Turing himself, from what I can tell (although remember this is not my area of expertise, I'm a clinical psychologist not an AI specialist) described the Turing test as a thought experiment game, not a test. In fact, there was no formal Turing test in his (distressingly short) lifetime.

As a clinical psychologist, however, I know that if you change expectations, you change experience. So a key issue with the current "win" is that it was not the software that fooled the interrogators, but the the way that the developers (and the rules of this particular interpretation of the Turing Test) set up the experience. By creating the back story of an eastern European boy with limited English skills, they change the psychology of the interrogators. They will interact with such a person differently: their expectations will change; the social conventions on how far to pursue a question will change; their tolerance for unusual, even unsatisfactory, answers will change. Don't believe me? Go watch a "brilliant" school play (I've seen a few absolute stunning performances) but then imagine seeing it at the Royal Theatre and ask if it would still be "brilliant".

By the way, I also have to assume that the judges don't know any 13 year old boys. Eugene Goostman is not even a pale imitation of the ones I know. Seems more like a precocious 8 year old with social difficulties.

Jaysen wrote:I guess the question of passing is predicated on what our definition of human interaction in a given medium is expected to be. If the test was state as "convince us that these txts come from a human", then my wife would fail.

But that's not the test (luckily for your wife and, for that matter, mine). Again, the exact nature of the Turing differs depending on who is presenting it (and what they get out of it) but in short it is that a human would not be able to tell the difference between a computer program and a live human being when each was trying to convince you they were human.

As for "the first" to pass the Turing test, well that depends on who's doing the defining. I've seen Eliza, that ancient parody of Rogerian nonspecific counselling, described as passing the Turing test. And it was developed in 1966! I've also seen mention of other chatbots in other so-called Turing Test competitions said to pass the test - they just didn't have as a good a media promoter as a certain academic at the University of Reading. :roll:

Jaysen wrote:My point is that academics forget about the real world too often. I think the test committee may have actually applied the following question to determine the "pass" for this test:
supposition wrote:Would this fool a normal person?

If they answered "yes" then it passed the test legitimately in my opinion.

Which means we are doomed on many different levels...


I'm not sure if any of the people I originally referenced were academics. As best as I can recall they included industry specialists, tech journalists and crank watchers, although there is one that I have read since that is an academic (see link below).

As to "Would it fool a normal person?", I still want to know under what conditions it would fool them. There's a difference between 2 or 3 exchanges and 10-15 exchanges and unlimited exchanges (the latter seems more like what Turing described). It's a lot more impressive to fool someone over a longer time period. One of the complaints against this particular Turing test was that the time for the exchanges was restricted, making it easier for the challengers - not what Alan Turing described in his thought experiment.

This is not to denigrate or dismiss the achievements of the developers of the Eugene Goostman chatbot. I'm just not sure that this is a sufficiently robust example of a computer fooling humanity into thinking it was one of us. Nice parlour trick, but not groundbreaking science ushering in a new era of our electronic overlords. We'll need to wait a little longer for SkyNet.

For a nice description of the issues, check out Scott Aaronson's blog post. Scott also quotes one of the developers of Eugene Goostman, Vladimir Veselov, acknowledging that the game was in their favour and hosing down claims that this is some great breakthrough.

And for a (supposedly) leaked image of AmberV responding to the Eugene Goostman furore, check out this. I make no claims regarding the veracity the image...
Complete and utter NOMsense.
Image

User avatar
Jaysen
Posts: 6294
Joined: Mon Dec 17, 2007 4:00 am
Platform: Mac + Windows
Location: East-Be-Jesus-Nowhere SC, USA

Fri Jun 13, 2014 1:32 pm Post

Valid points.

I do agree that there is much hype around this.

I think the point we are both making is that there needs to be a defined criteria to the Turing concept if there will be public claims of a "passing" of the test.
Jaysen

I have a wife and 2 kids that I can only attribute to a wiggle, a giggle, and the realization that she was out of my league so I might as well be happy with her as a friend. 26 years marriage later, I can't imagine life without her. -Me 10/7/09

ImageImage

User avatar
vic-k
Posts: 7135
Joined: Fri Apr 27, 2007 8:23 am
Platform: Mac + Windows
Location: Protesting in the nude, outside ex Red Lion TESCO Store

Thu Jun 19, 2014 3:29 pm Post

Can't understand all this fanny waffling, about something as simple and straight forward as a 'Touring Test'

"Contestant, please be seated."
"Thank you."
"Are you ready?"
"I am."
"Then we'll begin."
Q.1 Did you enjoy the tour?"
A. "Yes."
Q.2 "Would you go again?"
A. "Yes."
Q.3 "Would you recommend the tour to others?"
A. "Yes."

The Answer to the Touring Test, is: YES

Conversely:

Q.1 Did you enjoy the tour?"
A. "I */:?!$\ did not! It was a */:?!$\ load o' shite!"
Q.2 "Would you go again?"
A. No */:?!$\ way, pal! No Way!"
Q.3 "Would you recommend the tour to others?"
A. "I'd rather suck a bag o' */:?!$\ lemons!"

The Answer to the Touring Test, is: NO

So... What's the problem? :?
Vic
As a professional, you, are your one and only asset. Without integrity you are worthless, but with it, you are priceless.

User avatar
Jaysen
Posts: 6294
Joined: Mon Dec 17, 2007 4:00 am
Platform: Mac + Windows
Location: East-Be-Jesus-Nowhere SC, USA

Thu Jun 19, 2014 3:37 pm Post

Mr K, your solutions to life's complicated problems are refreshingly simple. Care to tackle a much easier problem? If so I offer the following stumper...
stumper wrote: How can I figure out what the wife really means?
Jaysen

I have a wife and 2 kids that I can only attribute to a wiggle, a giggle, and the realization that she was out of my league so I might as well be happy with her as a friend. 26 years marriage later, I can't imagine life without her. -Me 10/7/09

ImageImage

User avatar
vic-k
Posts: 7135
Joined: Fri Apr 27, 2007 8:23 am
Platform: Mac + Windows
Location: Protesting in the nude, outside ex Red Lion TESCO Store

Thu Jun 19, 2014 4:50 pm Post

Jaysen wrote:Mr K, your solutions to life's complicated problems are refreshingly simple. Care to tackle a much easier problem? If so I offer the following stumper...
stumper wrote: How can I figure out what the wife really means?

The answer to that question, would, I'm afraid, go right over your head (or stump, in this case). To figure out what a man's wife, actually, means, involves a process of elimination and deductive reasoning, a grasp of quantum mechanics, and a keen sense of the absurd. A tortuous path at best, and at worst a Freudianesque nightmare.

My advice to you, is, just hazard a guess, and run with that. No sense beating yourself up, striving for an unattainable goal.
Dr. Mulality.
As a professional, you, are your one and only asset. Without integrity you are worthless, but with it, you are priceless.