The role of artificial intelligence in preventing university dropout: a pedagogical reflection

Artificial intelligence (AI) offers new opportunities to tackle the phenomenon of university dropout, with important pedagogical implications. This paper presents a critical pedagogical reflection based on an experimental project launched in 2017/2018 at a university in Northern Italy, aimed at developing predictive software based on machine learning models (LDA, SVM, Random Forest) and a dataset of over 10,000 first-year students. Integrating socio-demographic, academic and psycho-educational variables has improved the system's ability to identify at-risk profiles early, thus supporting the development of personalised interventions. Dropout prevention is interpreted here as an opportunity to promote an educational model focused on the development of individual skills and the adoption of predictive models which, when properly integrated into a structured decision-making context, can become fundamental tools for more equitable, proactive and effective university governance in the fight against dropout.