Another important artificial intelligence tool is a genetic algorithm. A genetic algorithm is an artificial intelligence system that mimics the evolutionary, survival-of-the-fittest process to generate increasingly better solutions to a problem. Organizations apply genetic algorithms to problems for which (1) there are literally millions of possible solutions and (2) there is no appropriate problem-solving algorithm that will generate the perfect solution.

Consider the development of a new home area. Assume that someone has purchased a large parcel of land and intends to build 5,000 homes for residential sale. The task of laying out the design of the new home area is a daunting one. This is further complicated when having to determine how to optimally lay utility lines (fresh water lines, sewer lines, electrical lines, phone lines, gas lines, and cable TV lines). There are extensive sets of constraints concerning the laying of utility lines, but no simple algorithm that will help you find the right solution.

For example, depending on the grade of the terrain, fresh water and sewer lines must be sized to accommodate the slope of the terrain. And you can’t run electrical or gas lines under a bridge. Furthermore, cable TV substations have to be positioned appropriately to accommodate the use of high-speed cable TV modems. Those are just a few of the many considerations, never mind the most basic one of optimizing lot tracts to build the most homes while minimizing unused land.

Many residential home developers turn to a genetic algorithm for this problem. Once the genetic algorithm has been fed all the necessary information (maximum and minimum lot tract sizes, topographical information of the terrain, performance criteria for evaluating solutions, and so on), it sets out creating solutions. As it comes across a fairly good solution, it attempts to alter the solution (using evolutionary, survival-of-the-fittest processes such as crossover, selection, and mutation) to create a better one. Eventually, the genetic algorithm exhausts all possible solutions and makes a recommendation concerning the best one, given the constraints and performance criteria that were specified.

Of course, the decision is still yours to make. While a genetic algorithm can determine the best possible solution, you need to carefully analyze the solution and determine if it really is “best.” Technology can make recommendations, but the ultimate decision is still yours. In the end, you receive either the credit for making a good decision or the reprimand for making a bad one.