Abstract:
Artificial intelligence (AI) efficiently builds predictive models in engineering, politics, economics, and science, as well as provides optimal strategies for solving various problems. However, modern AIs are often far from being as accurate as one might have expected a few decades ago. As a result, a number of problems linked to the widespread use of AI hinder the positive effects of the tasks it solves. This article focuses on the difficulties and limitations in using AI systems that have arisen to date and possible ways to overcome them.