I suggest to make a simple prototype of AI based on the famous John Koza's Genetic Algorithm with some minor modifications. It will demonstrate and educate, those who are interested, in capabilities of that approach to program AI for automatic solution of problems in the "Tetris" puzzle game for example.
Target platform : Tablet Category
: Education (demonstration with explanations)
There is a web-site of Prof. Koza, where programming with a classic GA is explained:
I like to describe AI as some logical scheme, which consists of:
1) a node named "Sensors"; it gives information to the scheme;
2) a collection of nodes named "Angels"; each of those nodes is intended to analyze information from "Sensors" and memory to make its forecasts for any given plan of actions;
3) a node named "Judge"; it is intended to evaluate adequacy of each of the "Angels" by comparing their forecasts with reality;
4) a collection of nodes named "Souls"; each of those nodes is intended to analyze information from "Sensors" and memory to make its independent plan of optimal actions;
5) a node named "Desire"; it is intended to evaluate desirability of each plan from all of the "Souls";
6) a node named "Will"; it is intended to choose from all of the plans, which are evaluated by "Desire", only the best one, which commands executors, till a new better plan arises;
7) a node named "Actuators"; it gives information to executors;
8, 9) nodes intended to exercise genetical selection of the "Angels" and "Souls".
See the picture here:
If you are interested, please, contact me, because I need your help.