Дослідження методів штучного інтелекту для створення інтелектуальних ігрових агентів

The constant development of machine learning technologies has had a great impact on the gaming industry. Artificial intelligence allows the creation of increasingly complex and intelligent bots that are capable of learning and self-improvement, opening up new perspectives for game developers that allow them to create more interactive and immersive game worlds. This research aimed at the creation of an intelligent chess agent that, with the help of machine learning methods and search algorithms, would be able to independently improve its chess skills and adapt to the various playing styles of its opponents. Presented in the article information is a scientific and methodical character. It is expected that considered algorithms can be used not only in chess, but also in other fields that require decision-making under conditions of uncertainty. Various machine learning algorithms have been investigated, including traditional methods and deep learning methods (in particular, convolutional and recurrent neural networks). According to the analysis results, it was determined that deep learning is the most promising method for recognizing complex chess patterns and automatically forming effective strategies. DeepMind's AlphaZero artificial intelligence system, which is based on the principles of deep learning, was chosen in order to optimise the bot's decision-making process. AlphaZero is a powerful chess tool due to its self-learning ability, deep neural networks, and efficient moves tree search algorithm. It opens new horizons in the field of artificial intelligence and shows that computers can outperform humans in complex intellectual tasks, but also do it in fundamentally new ways. AlphaZero is not designed particularly for playing chess, it can be adapted for any game where the moves of the players alternate. It makes this engine a truly universal tool for artificial intelligence research.