On 23 February 2026, the AI Society at the University of Padua hosted a seminar by Dr. Eugenio Tufino (University of Modena and Reggio Emilia), exploring innovative approaches to designing AI tutors for physics education.

The seminar focused on role engineering—the practice of configuring large language models to adopt specific pedagogical behaviors. Through practical examples based on Gemini Gems, participants learned how generative AI can be designed to act as a tutor that encourages Socratic dialogue, metacognitive reflection, and independent problem solving, rather than simply providing answers.

Particular attention was devoted to the design of effective tutoring strategies, including the use of guiding questions, structured feedback, and explicit reasoning steps to promote deeper student understanding.

The seminar also addressed the current limitations of AI technologies, emphasizing the importance of human oversight in education and discussing emerging approaches such as Retrieval-Augmented Generation (RAG) for building specialized AI systems to support scientific reasoning.

Building on the Fall School and previous faculty workshops, this seminar provided participants with an advanced perspective on the future of AI-assisted learning environments in physics education.