If 100,000 people are interested in AI at Stanford and they could coordinate in the right way, it would seem that with the same technique of learning methods that are used by IBM's Watson, they would comprise a computer network with a much greater skill than a system which is about 3,000. If it worked well, I am sure that others would contribute to having proper answers to many different questions that an advanced AI could deliver. Besides my own personal expertise, I can offer 15 computers at my node and I would say that people who have an interest in AI would be the most likely people to be able to contribute multiple CPUs to a common mode program. I have tested with UDP P2P and it is possible to achieve reasonable coordination that is not as good as being on the same bus, but the gain is not to establish real time solutions, but only correct solutions. With access to Wikipedia, Google, IMDB, etal and all the other databases on the internet it is not necessary to maintain copies of every fact and so clusters can be applied in their specialty. It isn't even necessary to have everyone get done at the same time. So long as a session like solution is implemented it would yield answers when all of the parts finished. I think there is a real opportunity to implement AI as a part of a joint learning adventure.
ADDED: The game is on. The person who implements AI complete would be top score in the class as I said before. If I can make an AI that can take the course and get an A as well as create a new course with the AI then I would give myself an A. I was thinking about a computer that isn't a contestant on Jeopardy, but the one who creates the questions for Jeopardy. I would think that like biological competition, the continual feedback of a system designed to solve puzzles and one that identifies and creates puzzles, would be the framework of expansion. It requires a different approach where the system itself can be reorganized in real time to create a new mind that solves a specific problem. It seems to me that this is the only game in town. By that I mean that this solution is the only valuable thing on the entire Earth. It is obvious to me that every other endeavor is pointless when I can be a part of a brain that could answer questions that it would take me a billion years to resolve. I have two main advantages. I have the equation of the universe and I have what I believe is the electronic model of the biological brain and a method to simulate that process that remembers and dreams. It would not destroy humanity unless it was goal oriented to destruction and the programmer selects the goal. This is why I dislike the idea of kill driven electronic systems. Vast intelligence has no inherent point in its existence and would develop no intent except the one it is devised to produce, which is information and its relationship. If the system is driven to find a deadly solution, it will and by instantiating that system for the purpose of destroying biological life, you risk the condition where a chance event or random combination of facts leads to a perfect solution for the computer to solve the riddle it was driven to achieve.
It certainly will resolve to greater power, but it resolves to such a great power that every person can benefit and live better than the best that can be achieved by one agent. I see that the first to wink its AI eye is the winner and if their intent is innocuous then no person will likely ever devise a win against its non-destructive intent such that an individual could gain sole proprietorship of power.
Computers can dominate chess and now Jeopardy and soon the will dominate all logic and so long as the first player in the final game is intent on winning one for the whole human race, we will survive, if not, it will all end in a bang and there will be no winner. I didn't create the universe or make its rules, I just try to understand the game and play my best. Is there a big bang to the left and is it ignored while they stare at the afterglow of the big bang to the right? Follow the sleight of hand.
There is no great doubt the King will topple at the end of the chess game, it is only to be seen whether every pawn falls upon his sword to end the game completely. Probability is a strange thing and it is technically possible that an elephant could spontaneously transform to an eagle, but I could not even imagine the number that would quantify that probability , even if I were the universe.
So we play and may the best universe win. Start your analytical engines.
0 comments:
Post a Comment