Creation of Text-Based Training Materials for Personnel Training in Nuclear Power Enterprises Using Artificial Intelligence

The possibility of using generative artificial intelligence technologies based on large language models to create training materials for personnel training in the energy sector, using nuclear power enterprises as an example, has been researched. The relevance of the issue is deter-mined by the significant volume of regulatory and technical documentation, as well as the high resource intensity of the traditional process of preparing training materials. The regulatory re-quirements for personnel training, the current state of automated text summarization technolo-gies, and the risks associated with artificial intelligence hallucinations were analyzed. An ap-proach to automating the creation of the learning component of knowledge assessment courses is proposed by structuring, extractive and abstractive summarization of regulatory documents, as well as processing graphic materials included in such documents. An experimental test was conducted using common generative artificial intelligence tools, and the results were compared with materials created by experts in the subject area. The results show that comparable accura-cy and clarity were achieved without the negative impact of hallucinations and with a signifi-cant reduction in development time, provided that expert control by humans is maintained. We concluded that the proposed approach is practical and promising for improving the effective-ness of training personnel at energy companies.