Creating methodological support for a math lesson using neural network technologies

this article examines the potential of neural networks for the generation of independent mathematical tasks, specifically designed for students in middle and high school. Generative neural networks, such as: ChatGPT, YandexGPT, GigaChat, and MathGPT. The neural networks were provided with the same instructions, yet each demonstrated distinctive characteristics, resulting in variants that differed from one another. The article discusses the development of tasks of varying degrees of difficulty on a range of topics, including: The subject matter encompassed «quadratic equations», «Pythagorean theorem», and «limit of a function». The tests were evaluated by mathematics teachers through a focus group. In addition to assessing the tasks themselves, the teachers also evaluated which of the neural networks demonstrated the greatest proficiency in addressing the tasks, based on qualitative indicators and the applicability of the solutions in their respective classrooms. The experts highlighted that this approach to independent work is straightforward and accessible, and it significantly streamlines the preparation process for lessons. However, they also emphasized the need for task adaptation to ensure alignment with the specific learning objectives and instructional context.

generative neural networks, GPT, tests, independent work, mathematics teaching methodology, educational content, artificial intelligence, task generation