Generative AI has increasingly permeated higher education, offering new opportunities for personalised, inquiry-based, and reflective learning (Owoseni et al., 2024; Tariq, 2025; Yuan & Hu, 2024). However, its pedagogical implications for disciplines such as musicology remain underexplored. Grounded in Socratic pedagogy and the concept of deeper learning (Mehta & Fine, 2019), this study examines how the Socratic AI tutor Cogniti supports learning among first-year music students in the This is Music course at the Sydney Conservatorium of Music. Using an exploratory mixed-methods design, the study combines thematic analysis of 20 student-AI conversation transcripts with analysis of student feedback questionnaires. The findings indicate that Cogniti can effectively connect lecture content with course learning goals while providing multifaceted academic guidance. Indicators of deeper learning were reflected in students’ critical reflection, conceptual engagement with lecture materials, and their ability to connect academic ideas with personal musical expression. Questionnaire responses further suggest that most students perceive Cogniti’s problem-based questioning as stimulating reflective thinking and intellectual engagement. However, some students report practical challenges, including system downtime, conversational imbalance, and concerns regarding citation accuracy. Given the study’s exploratory nature and limited sample size, the findings should be interpreted as context-specific. Overall, the study highlights the potential of Socratic AI tutoring to support critical thinking in musicology education while raising important considerations for its responsible pedagogical implementation.