A national decision database. That''s my idea, it is basically a large search engine. When people have a problem they type in their problem and it matches it to similar problems. So you enter a problem like "Should I start smokeing?" and then it searches for the closest match to that; this example would probably get an exact match. The user then selects the closest match and it shows potential decisions for the problem, in this case it would be yes or no. Ok here is where the AI comes in, it would use a simple AI like weighting system to determine the best decision. Each decision starts with an equal value, it doesn''t really matter what it is so I''ll use 1. So yes=1 and no=1. Then users who have made the decision before come in and do a little vote type thingy. They say which decision they picked in their life and they say how much that decision benifited or harmed their life. So they say yes or no and something like "Great benefit, benefit, some benefit, no benefit and no harm, some harm, harm, great harm". Each one of these gets a weight, like Great benefit''s weight is 0.8 benefit is 0.5 some benefit is 0.2 neither is 0 some harm is -0.2 and so on. At the beginning both decisions had the same value of 1; now we take which decision the person picked (yes or no) and we multiply the value of that decision times the benefit or harm weight that they picked, and then add that value to the value of the decision. So if your decision was yes and you said it caused harm to your life you would have (1*-0.5)+1 and the yes value would become 0.5 and if another user said it caused great harm it would then be (0.5*-0.8)+0.5 which would be 0.1 . Then to determine which decision is better based on the user input the program just sees which has the largest value, in this case yes=0.1 and no=1 so the best decision would be no. After a while you wouldn''t have to make any decision on your own anymore, you could just go on the internet and have your computer tell you what you should do. hmm this post is kindof long and not written very well but if you get the gist of my idea tell me what you think.