ABSTRACT
The idea of banding together to resolve a problem is not limited to evolutionary algorithms. Another Artificial Intelligence method based on this principle and inspired by nature is multi-agent systems. Unlike evolutionary algorithms, they work directly in the state space. A computer program written in this manner, with basic programs coupled together, is called a multi-agent system. This type of system is somewhat similar to ant colonies: each ant does a very specific task that it repeats tirelessly for the colony to function as a whole. Each ant’s work appears to be very simple and, yet, when combined together, they produce a structure that functions in a very complex manner. To design a multi-agent system, people have to define what the different agents do, how they exchange information and how they work together to resolve a problem. Artificial intelligence takes inspiration from this technique to write algorithms capable of resolving difficult problems, such as the traveling salesman problem.
