Challenge Team Abstract
Team Number: 006
School Name: Albuquerque Academy
Area of Science: Artificial Intelligence
Project Title: Genetic Programming of Hive Intelligences
Genetic programming is a technique which allows one to teach a computer how to automatically program itself to accomplish a task instead of the programmer explicitly telling the computer what steps are necessary to complete this task. In order to do this, one provides a simple programming language which can be interpreted at runtime and contains commands which would be useful in computations related to solving the problem which you want the computer to solve. One then randomly generates a small init ial population of algorithms in this language and tests their ability to perform the desired task. The programs which come the closest to a solution are considered the most fit. These programs are then combined to create a new generation of programs, and the process repeats.
We intend to use these techniques to evolve a hive intelligence similar to those found in bees, ants, and other insects. Though each member in such an intelligence possesses only a small amuont of information, a hive of bees still manages to produce honey. Though no bee actually understands the process of producing honey, through a combined effort they are able to do so. We intend to evolve small computer programs that work together to accomplish a task, although none of the small programs will ind ividually understand how to complete the task.