Areas of application of artificial intelligence

What possibilities does artificial intelligence offer today? The short answer to this question is difficult to formulate, because in this scientific direction there are too many subdomains, in which a lot of research is carried out. Only a few applications are listed below as examples.

• Offline planning and scheduling. The NASA Remote Agent program, which operates hundreds of millions of kilometers from Earth, is the first on-board autonomous scheduling program designed to manage the scheduling of operations for a spacecraft. The Remote Agent program developed plans based on high-level targets set from the Earth, and also monitored the operation of the spacecraft during the execution of plans: it detected, diagnosed, and fixed problems as they occurred.

• Conducting games. IBM's Deep Blue program was the first computer program that managed to beat a world champion in a chess game after she beat Garry Kasparov with a score of 3.5: 2.5 in a demonstration match. Kasparov said that he felt in front of himself at the chessboard the presence of "intelligence of a new type." Newsweek magazine described this match under the heading "The Last Defensive Line of the Brain." The value of IBM shares rose by 18 billion dollars.

• Autonomous control. Alvinn's computer vision system was trained to drive a car, keeping to a certain lane. At CMU, this system was housed in a minibus operated by the NavLab computer and used to travel around the United States; over 2,850 miles, the system provided steering for 98% of the time. A person took control only for the remaining 2%, mainly on exit ramps. The NavLab computer was equipped with video cameras that transmitted images of the road to the Alvinn system, and then this system calculated the best direction of travel based on the experience gained in previous training runs.

• Diagnostics. Medical diagnostic programs based on probabilistic analysis have managed to reach the level of an experienced doctor in several areas of medicine. Heckerman described a case where a leading expert in the field of lymph node pathology did not agree with the diagnosis of the program in a particularly difficult case. The creators of the program suggested that this doctor ask the computer for an explanation of this diagnosis. The car indicated the main factors that influenced its decision, and explained the nuances of the interaction of several symptoms observed in this case. In the end, the expert agreed with the decision of the program.

• Supply planning. During the Persian Gulf crisis in 1991, the DART (Dynamic Analysis and Replanning) system was deployed in the US Army to provide automated supply planning and scheduling of shipments. The operation of this system covered up to 50,000 vehicles, units of cargo and people simultaneously; it had to take into account the points of departure and destination, routes, as well as eliminate conflicts between all parameters. Planning methods based on artificial intelligence made it possible to develop such plans in a matter of hours, which would take weeks to compile using the old methods. Representatives of the Agency DARPA (Defense Advanced Research Project Agency - Directorate for Advanced Research Programs) said that this application alone paid off a thirty-year investment in artificial intelligence made by this agency.

• Robotics. Many surgeons now use robotic assistants in microsurgery. For example, HipNav is a system in which computer vision techniques are used to create a three-dimensional model of the anatomy of the patient’s internal organs, and then robotic control is used to guide the process of inserting a prosthesis that replaces the hip joint.

• Understanding of natural language and problem solving. Proverb is a computer program that solves crossword puzzles much better than most people; It uses constraints that determine the composition of possible word fillers, a large database of previously encountered crossword puzzles, and many different sources of information, including dictionaries and operational databases, such as movie lists and actors who played in these films. For example, this program is able to determine that one of the solutions that are suitable for the “Nice Story” key is the word “ETAGE”, because its database contains a key pair - the solution “Story in France / ETAGE”, and the program itself recognizes that Nice X and X in France often have the same solution. The program does not know that Nice (Nice) - a city in France, but is able to solve this puzzle.

The above are just a few examples of artificial intelligence systems that currently exist. Artificial intelligence is not magic and not science fiction, but a fusion of methods of science, technology and mathematics.