Exactly. The ability of the computer to comprehend and actualize responses from natural language. That is the main missing factor (there are others, but this one is the most obvious). For one to be able to have a ‘conversation’ with a computer, and not be able to tell the difference. I would be quite interested to watch this matchup, in particular if the questions that are asked are asked in the normal Jeopary manner. That will be quite the test. Someone asked if it is just someone googling responses, but that would be too slow against someone like Jennings (taking out the part of a human googling queries); and in the same breadth, even though a computer could do it much faster, it would have to understand natural language. Not set commands or key inputs ...normal natural language. The moment a computer can do that, and do that in a manner that is indistinguishable from an actual human being, is a point in time when computing will have taken an interesting (and quantum) jump. This, if it works, is quite a huge event.
“Exactly. The ability of the computer to comprehend and actualize responses from natural language. That is the main missing factor (there are others, but this one is the most obvious). For one to be able to have a ‘conversation’ with a computer, and not be able to tell the difference. I would be quite interested to watch this matchup, in particular if the questions that are asked are asked in the normal Jeopary manner. That will be quite the test. Someone asked if it is just someone googling responses, but that would be too slow against someone like Jennings (taking out the part of a human googling queries); and in the same breadth, even though a computer could do it much faster, it would have to understand natural language. Not set commands or key inputs ...normal natural language. The moment a computer can do that, and do that in a manner that is indistinguishable from an actual human being, is a point in time when computing will have taken an interesting (and quantum) jump. This, if it works, is quite a huge event.”
This is a LOT harder than folks on this thread are giving credit for. However, I don’t think it is quite as revolutionary as that. The only thing that makes it feasible is the very rigid format of jeopardy. There is a subject and an answer. So key meaning parsings have been done for the computer already. That’s a long way from being able to have a conversation about an imaginary blue pony friend named Pony with a five year old while driving.
Text mining applications have only a decent success rate in extracting a subject from text and then a sense whether the text is positive or negative about the subject. They miss a lot.
If you are interested in this, one interesting and ambitious effort related to this is CYC from MIT. It has been in development for a decade and attempts to encode common sense for a computer. Link follows:
http://en.wikipedia.org/wiki/Cyc
When this becomes reality many people on the left-hand side of the IQ bell curve will become more or less useless from the point of view of the economy. I think we are already observing that the productivity gains of the last few decades resulting in growing chronic unemployment.