@Sorian IBM have released the API so people can use the Watson AI's powerful analytics processing power for what ever purpose they desire, I was wondering , would it be possible to pass Nerual Network (NN) data to watson so watson can make decisions and weight the NN decision outcomes?? And do you think Watson's advanced natural prediction and somewhat human styled thinking ability might make a difference in the way AI might work if there was a way to pass it NN data and have it analyze it and send back weighted responses
I have seen something similar being used to teach a robot to move by it'self. I'll see if i can find the video you might find it interesting. I don't know if that kind of learning would benefit a strategy game, i think a more pre-programmed set of strategies is more effective if the AI correctly chooses the best strategy for each situation as opposed to blind learning. My biggest thought on using Watson to do the Neural Net calculations was to offload some of the server processing requirements to do this for multiple groups of units to another server dedicated to analyzing data and making optimal decisions on which outcomes would be best to use. Sort of like outputting the neural network data to a chessmaster and asking him for the best advice on the next move saving you from doing all the thinking yourself. Kind of a way to improve server performance when heavy AI calculations or multiple calculations are required.
