Data mining and soft computing technologies, make up most of the space in Computational Finance. The models 'use' Predictive intelligence and increase reliability of applications as compared to statistical results in academic research and theory. The use of program trading, is discussed in my paper on RG. The applications include Trading, Credit risk modelling, default classification, customer segmentation and many others employing Genetic Algorithms, Neural Networks, Fuzzy Logic, and various algorithms such as PSO, ACO, BCO, and NIA in general.
You cant really make use of AI in Economics/Finance as the domain is dominated by belief (memes) . Any model that works now will not be guaranteed to work later and worse still at some point even later it will work again. Some heuristics will improve your odds, but usually only during quite times - currently the global finance experiments are moving into new territory so very little prediction is possible. AI needs to sample a consistent environment to work.