Abstract:

Integrating generative artificial intelligence (GAI) into learning ecosystems is one of the major challenges currently facing higher education and requires a thorough review of the teaching methodologies employed in the classroom. The activity described here is embedded in a second-year course of a Bachelor’s Degree in Physics. The course aims to provide students with basic skills in metrology and electronic instrumentation for the measurement of physical quantities. In this context, students are introduced for the first time to basic concepts in electronics and, in particular, circuit theory.

This proposal explores the use of a GAI-based tutoring system, while ensuring quality control and appropriate timing of the responses delivered. Identifying concepts that students find difficult to understand is a key element of a student-centered active learning and additionally enables the development of specific instructional interventions. A key pedagogical challenge lies in facilitating access to such interventions at the time and in the manner in which they are most relevant. Academic tutoring is a valuable tool for this purpose. Its mission should be to individualize and personalize the learning process for each student, helping them to expand and deepen information, resolve doubts and overcome difficulties. Continuous tutoring by the instructor makes it possible to determine the degree of success in the various assigned tasks, adjust initial objectives when necessary and guide autonomous learning. In the current technological landscape—characterized by the widespread use of GAI tools and by the difficulties involved in verifying the authorship of generated responses—an update of this tutoring model is required.

The activity consists of students collaboratively analyzing GAI-generated answers to a trial test, thus enabling reflection on the accuracy of the initial responses and an assessment of their improvement as the wording of prompts is refined. In this verification process, the questions designed by the instructor are intentionally constructed to encourage the AI to produce ambiguous or imprecise answers. In the classroom, students must query a GAI tool and, working in groups and with the instructor’s guidance, assess the suitability and clarity of the generated responses, determining whether each answer is valid and/or sufficiently precise. Groups are also required to formulate queries using different prompts in order to develop competence in prompt design, analyzing and comparing the resulting sets of responses according to the context provided to the tool. Finally, a refinement process will be carried out, if necessary, until convergence is reached on the bounded answer established as valid by the instructor.

This activity is designed with a dual educational objective. On the one hand, from an academic perspective, it aims to support students in deepening and improving their understanding of key concepts. On the other hand, it seeks to foster critical thinking and the responsible use of GAI tools, including appropriate contextualization in prompts, precision, comprehension and verification of responses, time invested in the process, responsible use of these tools and ethical considerations. Overall, students are encouraged to recognize that, while GAI outputs are easy to generate, they are often difficult to validate and therefore require a reflective and informed approach to their use within academic learning processes.

Keywords:

Generative Artificial Intelligence, Student-centred active learning, Academic tutoring, Critical thinking.