The phrase “artificial intelligence system” for many causes associations with various science fiction films and interlocutor programs that emulate artificial intelligence. Robots have become a reality in our time, and every time you open the next exhibition devoted to robotics, you are surprised at how far humanity has advanced in its technological progress.
The problem of artificial intelligence is due to the fact that, according to generally accepted ideas, man-made intelligence is such a computer process, the properties of which are associated with human thinking. However, science still cannot figure out exactly how a person thinks and what his thinking is. Therefore, the
creation of artificial intelligence is based so far only on intuitive guesses.
Meanwhile, one of the most promising directions in the development of modern information technologies has been the creation of applied neural networks. What is an artificial neural network (ANN)? This is a small mathematical model that works on the principle of biological neurons that are functionally integrated into a single system.
Man-made
neural networks, or, as they are also called, artificial intelligence systems, are often used to search for solutions to problems with an incomplete number of data or tasks that cannot be clearly formalized.
The first ANN appeared in 1958 thanks to the psychologist Frank Rosenblatt. This image-based system simulated the process of the human brain and made attempts to recognize visual data. The principle of operation of the ANN is based on creating a connection between the totality of the processed elements. At the input, each neuron receives a large number of signals. He performs their analysis in accordance with weighted coefficients and generates a personal signal arriving at another neuron. All neurons are organized into layers and are connected to each other. Each layer processes the input signal and then forms its own for the next layer. The main advantage of ANN is the ability to self-learn.
For the operation of the artificial intelligence system, it is desirable to use several processors, since when using only one computer, the speed of operation decreases noticeably. Such ANNs are used for the synthesis and recognition of speech, handwritten text, in the field of finance, as well as wherever it becomes necessary to analyze powerful information flows.
Currently popular neuro-expert systems are special systems of artificial intelligence, the basis of which is a huge knowledge base. It stores numerous information and methods necessary to solve the tasks. The database also contains a self-learning algorithm that relies on the procedural data of decision estimates.
A very important component of any expert system is its interface. Thanks to him, a person can fill the database with new data, receive logical conclusions, etc. Using accumulated knowledge, these systems can find the right solution for those tasks that are too complex for human capabilities. Expert systems are often used in areas such as program design, military science, geology, planning, forecasting, medicine, and training.
Recently it became known that Google intends to provide search processing to a new artificial intelligence by 2029. Moreover, according to the words of technical director R. Kurzweil, a new intelligent search engine will be able to understand human emotions. Isn't that amazing? Robots do not know how to think, but can learn. And what will happen next?..