Abstract:

Artificial Intelligence in Education (AIED) is a popular research and application field, an so is Music Information Retrieval (MIR). When the two meet in class, intriguing and effective teaching and learning experiences emerge.

This paper focuses on a years-long experience of teaching and learning programming and AI using music datasets. It describes a model for creating such datasets for effective use in class, as well as how to evaluate the datasets created following this process. By presenting several examples of such datasets, the paper also demonstrates how they can be used for educational purposes and what pedagogical effects can be achieved that way.

The paper also goes one step further by discussing practical implications of such a teaching and learning approach. It suggests guidelines for creating assignments related to music datasets in teaching programming and AI and also discusses how ths approach can be expanded effectively. Limitations and alternatives of the approach are presented as well. The paper also stresses the benefits and downsides of using Large Language Models (LLMs) in teaching and learning scenarios leveraging MIR and AIED.

Keywords:

Music Information Retrieval, Artificial Intelligence in Education, Large Language Models, programming, AI, education, teaching, learning.