First computer has passed the Turing test

[Roxxo]

Looks like they beat Ray Kurzweil's prediction by quite a few years. He predicted that a computer would pass the Turing test sometime by 2029.

 

[edco76]

Looks like a lot of Sherdoggers may be getting a girlfriend sooner than expected.

 

[Raiman]

right.

the "AI" does not even know it is trying to be a person. it does not even understand what a person is. it is just something programmed by a person with certain responses and probabilities.

a glorified teddy ruxpin

I guess I think that it has only been a idea proposed for what 65 years or something and just like computers beating humans in chess it is kinda a big deal. Maybe it's not a sexbot that has thought and feelings; it's not a replicate from Bladerunner, but it is a milestone and the first step toward AI programing.

I look at it this way, how long did it take to go from a plane flown by the Wright brothers to prop engine planes in 1914 to jet engines etc. In other words - So big deal you could fly a few hundred feet to transcontinental flights and space rockets. Just my opinion.

 

[WARRIOR POET]

I guess I think that it has only been a idea proposed for what 65 years or something and just like computers beating humans in chess it is kinda a big deal. Maybe it's not a sexbot that has thought and feelings; it's not a replicate from Bladerunner, but it is a milestone and the first step toward AI programing.

I look at it this way, how long did it take to go from a plane flown by the Wright brothers to prop engine planes in 1914 to jet engines etc. In other words - So big deal you could fly a few hundred feet to transcontinental flights and space rockets. Just my opinion.

right. but the difference between this AI and the AI we think about in sci-fi movies is like the difference between a space shuttle and The Event Horizon

i might be limited, but i just don't see how we will ever get to an AI truly capable of understanding the world.

i think a Terminator like the 1st movie will be as far as we ever get to it.

Sarah Connah!

 

[flanellograf]

It has nothing to do with that. It's the philosophical argument that one can't distinguish between somebody who can speak with you in a language, and one who truly understands it. Somebody could whisper in my ear, or tell me what to type in German for example, and it could fool a German speaker, into thinking I actually understand the language. But I don't. I am only following instructions. I could pass as a German speaker on this forum, if I hired a German to give me instructions. There would be no way to tell if I actually understand the language, or am just following a program.

Yes, that's the basic form of the argument. I was talking generally, adressing the crux of all arguments in the same vein. Sorry if that wasn't clear.

 

[flanellograf]

There are literally dozens of replies to Searle's argument, it's certainly not consensus between scientists and philosophers. I personally think he underestimates computational technology and overestimates the uniqueness of the human brain. We'll have to wait and see.

Well, he was right in the naive sense that if you program a computer with instructions what to do, like a regular application, it won't really "understand" anything. But it turns out that is extremely difficult to accomplish in the first place, and not really a feasible way of doing it. The sheer number of instructions would be astronomical, and even with statistical modeling it's very hard to accomplish. Otherwise we would have automatic language translation working already.
What he didn't envision however, was programmatically creating the building blocks and hierarchical structures for learning we see in the brain, and then teaching the software just like it were a child. There already is software and a company that has been pretty successful, albeit still only doing (for the brain) simple pattern recognition like recognizing melodies, what a dog looks like etc.

 

[flanellograf]

Looks like they beat Ray Kurzweil's prediction by quite a few years. He predicted that a computer would pass the Turing test sometime by 2029.

No, they didn't. Fooling 30% of humans that it's a 13 year old boy from Ukraine by using conversation-logic-instructions, does not equal "greater-than-human intelligence".

 

[Heefer Madness]

No, they didn't. Fooling 30% of humans that it's a 13 year old boy from Ukraine by using conversation-logic-instructions, does not equal "greater-than-human intelligence".

Turing test has nothing to do with a machine having "greater-than-human-intelligence". It's about a machine passing for a human in some circumstances (natural language conversation), which is exactly what it did.

 

[Diktaattori]

Looks like a lot of Sherdoggers may be getting a girlfriend sooner than expected.

 

[WARRIOR POET]

No, they didn't. Fooling 30% of humans that it's a 13 year old boy from Ukraine by using conversation-logic-instructions, does not equal "greater-than-human intelligence".

NAMBLA members be on the lookout

 

[WARRIOR POET]

Turing test has nothing to do with a machine having "greater-than-human-intelligence". It's about a machine passing for a human in some circumstances (natural language conversation), which is exactly what it did.

well, then my VHS was my sex box decades ago so i guess we are late to teh party

 

[weed]

Destroy it now before it learns and evolves.

too late, I already created a sherdog account.


wait, does that mean I devolved?

 

[Organic Damage]

I still don't get why the bar is being able to fool 33% of people.

What I want to know is by what percentage a typical 13 year old human passes the Turing Test, assuming he passes at all (which I guess given teenagers isn't guaranteed).

 

[Iroh]

Tricking humans isn't that hard. Trust me.

 

[flanellograf]

Turing test has nothing to do with a machine having "greater-than-human-intelligence". It's about a machine passing for a human in some circumstances (natural language conversation), which is exactly what it did.

Did you even read the post I replied to?

 

[Heefer Madness]

Did you even read the post I replied to?

Yes. He said the event happened earlier than Kurzweil predicted, which is correct. It has nothing to do with a computer being smarter than a human.

 

[flanellograf]

Yes. He said the event happened earlier than Kurzweil predicted, which is correct. It has nothing to do with a computer being smarter than a human.

Kurzweil predicted the singularity, which is what I was talking about. If he said this would be proven by the turing test, then he obviously was wrong.

 

[Heefer Madness]

It should be pretty clear from my post that I realize that as well. As I specifically adressed it. And Kurzweil predicted the singularity, which is what I was talking about. If he said this would be proven by the turing test, then he obviously was wrong.

http://en.wikipedia.org/wiki/Predictions_made_by_Ray_Kurzweil

So I'm supposed to guess that you were talking about the singularity when you argue with a guy talking about the Turing test. Alright.

 

[Heefer Madness]

Kurzweil predicted the singularity, which is what I was talking about. If he said this would be proven by the turing test, then he obviously was wrong.

I see you edited, and this version is even more embarrassing. Yeah, except he wasn't saying that at all.

 

[flanellograf]

http://en.wikipedia.org/wiki/Predictions_made_by_Ray_Kurzweil

So I'm supposed to guess that you were talking about the singularity when you argue with a guy talking about the Turing test. Alright.

If you know anything about Kurzweil, and read the first post in the thread, understood my post, and were not hell bent on being an aspie about it, yes.