Distributed artificial intelligence allows a number of different computers to work together to solve complex problems and make decisions independent of human operators. This type of artificial intelligence is relatively new and builds upon previous research and development that allows the creation of machines and programs that are able to analyze situations and to make decisions that allow them to achieve their goals. The types of decisions these programs are able to make are limited by the constraints of the program, however.
In order for machines to have distributed artificial intelligence, they must be networked, either physically or remotely. The networking of machines allows them to communicate with one another and to share resources. A network of these computers can be far more powerful and faster than a single computer could be because each of the computers in the network can be in solving the complex equations they use in the decision making process.
Every distributed artificial intelligence network is made up of a series of agents. Agents may be individual computers, programs running on a computer, or programs that operate across a number of different machines. They may also be connected to sensory equipment, such as cameras or thermometers. The agents in a distributed artificial intelligence network are able to make observations about their environment, whether physical or artificial, and decisions about which actions to take within that environment. When the intelligence of these agents is distributed among many of them, each individual agent might only be responsible for a small portion of what the network perceives and does, making it necessary for these agents to communicate with each other in order to achieve the goals set by human programmers.
Distributed artificial intelligence, in most cases, makes use of agents that perform different tasks. One agent may be devoted to planning future actions, while another may contain information about previous experiences. By bringing these agents together, a system can arise that is much more intelligent than any of the agents would be on their own.
In traditional artificial intelligence with only one agent, the intelligence of the agent is modeled after the intelligence of a single life form, such as an ant or a human being. The behavior of distributed artificial intelligence networks, however, is modeled after the behavior of ecosystems. Taking an individual agent out of the network is akin to removing a plant or animal from the ecosystem and can damage the effectiveness of the system but will not always cause the entire network to fail.