Artificial intelligence (AI) and expert systems are related in that the development of artificial intelligence is usually built upon a number of expert systems. Expert systems are software programs interacting with a database of information gathered by human experts with varying viewpoints, and inference engines to quantify and analyze it. In order for artificial intelligence and expert systems to work together seamlessly and mimic the capabilities of human thinking, they are often built upon an array of microprocessors. These processors work in parallel to analyze and compare stored and real-world data and come up with meaningful results in a reasonable amount of time.
A good example of artificial intelligence and expert systems at work is the Watson computer created by IBM® corporation in the United States, over the course of three years. Watson is an internally-networked computer system of 2,880 microprocessors and 16 Terabytes of RAM memory that processes 500 gigabytes of data per second to analyze human speech. That is the equivalent of being able to read and analyze 1,000,000 books every second. More than 100 different expert system techniques run on Watson to compile meaningful answers to questions. The system accesses data from encyclopedias, literature, and contemporary news articles, and uses neural networking and other adaptable expert system software methods to comprise a rudimentary artificial intelligence that finds meaning in human speech patterns.
AI programming can be built on a number of different design methodologies, however. General human intelligence AI systems, known as “strong AI,” are those that most heavily draw upon the need of multiple expert systems running in tandem. One of the methods for developing artificial intelligence and expert systems in this manner is the use of fuzzy logic programming, which is software that attempts to quantify the vague nature of the real world that humans are good at understanding, but digital computers are not. Fuzzy expert systems work well where machines must adjust to rapidly changing real world conditions, such as in automobile automatic transmissions, dishwashers, cameras, nuclear power plants, and so on. Computer intelligence in Japan has made far more use of fuzzy logic programming than elsewhere, which may account for the nation's ability to lead the marketplace in advanced AI robotics.
Expert systems, therefore, are a fundamental component of any functional AI. Combined, expert systems attempt to circumvent the roadblocks that traditional computers come up against where every decision must comprise a discrete yes/no, true/false response. They do this by dynamically processing queries instead of following a predetermined program path, and weighing the values of each potential answer against each other. Building artificial intelligence and expert systems using heuristics, or the form of trial-and-error analysis that humans use regularly on a one-on-one conditional basis, instead of merely applying specific stored knowledge, is the next generation of machine intelligence that has the capacity to grow and learn over time.
