Autonomous agents for information processing

In many information processing scenarios human beings are able to cope well up to a specific level of information throughput. Above this level some kind of overload occurs and the human being or team of workers is unable to cope and becomes chaotic.

Examples of such scenarios are the military intelligence cell and air traffic control.

In both these examples the problem is generally divisible into sub-problems. Ingenuity is needed in defining the way the problem is to be subdivided and also in defining the communication protocol to be used between the members of the team.

An effective recursive division of the problem allows semi-autonomous agents to take control in a way that is scalable. As the size of the problem increases it should be possible to recruit more agents without limit and so avoid the chaotic fragmentation characteristic of traditional teams of workers.

Neural Solutions has built a protoype distributed AI system using the Windows Dynamic Data Exchange protocol across a network of computers.