Daniel Ślęzak
Research objective: The aim of the article is to analyse the theoretical foundations and empirical evidence for the effectiveness of AI systems in adapting learning pathways to the individual needs of medical students, and to identify the barriers limiting the full use of this potential. Methodology: The study is based on a systematic review of the scientific literature indexed in PubMed, Scopus and Google Scholar, covering publications from 2013–2025. Meta-analyses, systematic reviews and empirical studies on intelligent tutoring systems, adaptive learning platforms and large language models were analysed, together with European Union legal acts governing the use of AI in education. Main conclusions: AI systems significantly improve the learning outcomes of medical students, and the effectiveness of intelligent tutoring systems approaches that of one-to-one human tutoring. Personalisation works well for theoretical knowledge and diagnostic competences, but remains limited in developing soft and clinical skills. Implementation is accompanied by ethical risks (data privacy, algorithmic bias, hallucinations in language models), legal requirements (AI Act, GDPR) and organisational constraints. Application of the study: The findings may be used by medical university authorities designing digital transformation strategies, by teaching staff introducing adaptive tools into their courses, and by accreditation bodies and regulators developing quality standards for educational AI systems. Originality/Novelty of the study: The article combines a review of international empirical evidence with an analysis of the European regulatory framework and formulates concrete recommendations for Polish medical universities, including participatory system design and the inclusion of AI Act requirements in accreditation procedures.