Intelligent scaffolding for musical practice: a systematic review of AI and digital support for the self-regulated learning cycle
Related Experiment Videos
Xiuran Li1,2, Mar Galera-Núñez2, Yao Zhang2
1School of Music and Dance, China West Normal University, Nanchong, China.
Introduction:
Independent music practice depends on learners' self-regulated learning (SRL), yet how digital tools and artificial intelligence (AI) support the forethought, performance, and self-reflection phases of the SRL cycle, and when such support may undermine learner autonomy, remains unclear.
Methods:
Following PRISMA 2020, we searched Web of Science, Scopus, ERIC, and PsycINFO through March 2026. Sixteen empirical studies met the inclusion criteria and were appraised with the Mixed Methods Appraisal Tool. Because of methodological heterogeneity, we conducted a thematic synthesis rather than a meta-analysis.
Results:
Three categories emerged: traditional digital tools, discriminative AI, and generative AI. Traditional tools supported planning and reflection; discriminative AI enabled real-time error detection; and generative AI may help learners manage task difficulty and translate expert feedback into actionable steps. Study quality was uneven, and several findings came from small-sample or cross-sectional designs, which limits causal inference. Two potential risks, over-scaffolding that may diminish autonomy and visual feedback that may divert attention from sound, were insufficiently studied.
Discussion:
Digital and AI tools appear to scaffold different phases of SRL, but learner autonomy should remain at the core. Fading explicit support and pedagogical guidance may help reduce dependence. Longitudinal studies are required to test these mechanisms.