One of the more prominently advocated advantages of Artificial Intelligence in the context of Music Education and education in general is its possibility of personalizing the learning experience. We agree with that assessment, but must note that this ideal goal has not been satisfactorily attained in many research efforts. Most personalized systems in literature only rely on learner models that allow teachers to track students’ progress and suggest adequate resources. Both the models and the resources are however fixed and finite, limiting the amount of personalization that is possible. We take music education in particular as a field that can greatly benefit from a higher grade of personalization, and consider the possibilities offered by AI generation for music. We posit that with the currently available music generation technologies it should be possible to provide students with exercises created ad hoc for them to address their specific needs, guided by their teacher’s indications and concerns. Such a system could obtain a far finer grain of personalization than what is currently available. In this article we review the literature to assess the current standing of musical technique education personalization, and what is possible given the current available technology. We then describe prototypes to show how exercise generation can effectively enhance the personalization of music education, and suggest directions for further research on the topic.
Carnovalini et al. (2026) studied this question.