PROBLEMS OF ARTIFICIAL INTELLIGENCE FORECASTING AND ITS CAPABILITIES

Atamanenko Vadim

With the development of artificial intelligence technologies, the number of different areas in which their implementation is possible is also growing. At the same time, despite the global development and implementation of artificial intelligence in various spheres of public relations, a number of practical problems related to the capabilities of artificial intelligence remain unresolved. Among such practical problems are the possibilities of artificial intelligence aimed at forecasting. The author of the study concludes that the current problems of predicting artificial intelligence at the moment are due to the specifics of training artificial neural networks underlying it, which does not allow artificial intelligence to predict events. However, despite the existence of theoretical and practical problems of predicting artificial intelligence, it seems possible to improve the predictive capabilities of artificial intelligence. First of all, it is necessary to develop a fundamentally approach to the training of artificial neural networks, focused on improving the prediction process by artificial intelligence using artificial neural network learning algorithms, focused on the use of so-called "collective intelligence" abilities in training. The main problem to be solved when using this approach is the limited possibilities of artificial intelligence training in terms of forming an experience of reaction to similar situations, which can be relied on, as well as the ability to recognize similar situations when making decisions and forecasting.

Keywords:forecasting, artificial intelligence, artificial neural networks, training of artificial neural networks, collective intelligence