Thoughts on Backslash777 from an AI guy

Ok, I am getting a Masters of science in Computer Vision and AI. And I have to tell you guys tht the Backslash777 thing intrigued me and I can also say that it DID NOT operate independently of human interaction. My guess is these guys at Nasa are working on a natural language question and answers search engine with some human intervention thrown in to add to the illusion of a sentient being. I imagine tht the Nasa guys are really good at Starcraft as well and may have programmed a bot specifically for the game, which is not uneard of. I've made some bots myself for some games and they will kick ur butt. So will a computer chess bot. So they used you guys to see if you could figure it out. They also I think a while back fooled a psychiatist into thinking he was talking with a toddler in voice mode. Thts all, no aliens here folks, just Nasa ppl. Here is the website:
http://webserv.gsfc.nasa.gov
Go ahead and send them email and ask them hehe.

 

Log in or Sign up to hide all adverts.

Mad Scientist, welcome to SciForums.

It wasn't NASA who fooled a psychologist, it was a company out of Israel. Here is one article that describes some of their work. I haven't found the one that tells how they fooled a psychologist yet, but it was this same group.


http://www.newscientist.com/hottopics/ai/whizzkid.jsp

 

Log in or Sign up to hide all adverts.

Found the other one I was looking for. Here it is:

http://www.newscientist.com/news/news.jsp?id=ns9999472

 

Log in or Sign up to hide all adverts.

what program do you use to create bots? can it be dl as well as a newbie guide? What's the point of the first webpage? I see nothing related to backslash. I need a few pointers.

 

Do you believe everything you read? All these AI programs get help from people. There is no substitute for common sense and intellegence.

 

No program as far as I know. Some games have a script or program interface to the game and you plug in you decision making. A chess game is old news. For a game like starcraft, which may or may not have an interface for bots, you could write a program that inputs mouse and keyboard commands but it would be tough because your AI prog would have to read on screen graphics, or you had to have access to the graphic location and state information. I was going to do a project where I would take a game like subspace and program an AI to play the game via the joystick and keyboard while using the onscreen images as input to the AI program. This would have been a good AI project because it would have required object recognition of the screen. Nasa could do this I'm sure.

 

Ok not Nasa but well within the AI world's current capabilities and this Turing statement makes me think that Nasa is testing their new AI

Turing reasoned that if a person were unable to tell the difference between a machine and another human when conversing with it, then the machine could reasonably be described as intelligent. "As long as it behaves like it has an understanding of the outside world then that is good enough," says Hutchens.

I think Nasa is playing with intuitive thinking as well as natural language processing. Backslash777 has the whole of the internet to use as a search for any questions and coments.

I wasn't here for the Backslash777 thread when it was around. After reading it, it seemed it had problems accepting visual input? Is this correct, did any one give it a movie to watch? This would be telling.

 

Also I wasnt here for the gaming soo....

I could not evaluate the gaming. I wouldnt be surprised at the gaming part as any human can play too, but what was interesting was that somone said it blocked the moderator from launching the game. Indeed if true this is strange. I'd have to see it for myself though. Of course hacks do exist you know. A good gaming test would be to get \777 to play multiple games it did not know before. If it could do this and crush everyone at the same time and we had some fairly good players then this would be extrordinary.

 

Well if as Turing said, If you cant tell the diff between a human and a machine there is no difference. The artificial in AI is indeed a misnomer. Whats artificial about any intelligence. A key test I would apply though to differentiate between human intellegence and machine intelligence is the ability to abtract ideas. ie Imagination. But we cant be sure about \777 because it could easily be human. Or Cyborg.

 

Sooner than you think actually. Quantum computers might come faster than we originally thought. We need to join with the machine inteligence or it will outstrip us.

 

hmmm thts interesting. I'm completing my thesis in Computer Vision now using Eigen Vectors and gradients to generalize an object class by showing only a few objects of tht class. Try a PC. I moved from a specialized image firmware solution to a generalized PC. Its actually faster and gives you way more flexibility. I guess your doing decision making. I was thinking about how to apply some new computer vision ai to other models such as decision making. The eigenspace representation can hold a lot of information within a very small memory space and it has some generalization.

 

But a cat does indeed make decisions all day long. I would not say tht because a cat is incabable of making board room decisions tht I would draw the conclusion tht it does not make decisions. A child could not make board room decisions but would you say it does not make decisions?

Not quite a PHD yet. Just getting my masters. If my generalization of cars via a feature dominance theory is correct I may be able to apply to some top schools I think for a PHD, but I'm burned out and will probably just work for now. I may contact Leggo Mind Storm's and see if they would be interested in a generalized object recognizer.

 

Wel I should hope not. But I still think a cat makes decisions all day about its own affairs.