Music students found AI useful, not a replacement for teachers

The audio task

A music educator or ed-tech developer deciding how much of a lesson to hand to an AI practice tool needs evidence about how students actually respond to features like voice interaction and instant feedback, not just whether the software works technically.

What the documents show

A study published in Frontiers in Education on 10 October 2025, by Jingyuan Tan, Clinton Chidibere Anyanwu, Yaoping Liu, and Sri Suryanti, surveyed 600 Chinese undergraduate music students and followed up with semi-structured interviews of roughly 33 to 35 of them. Using questionnaires built on the Technology Acceptance Model, a stimulus-organism-response framework, and an uncanny-valley scale, the authors report students rated voice interaction and immediate feedback highly useful for skill practice, while also describing discomfort with the “emotional element” of AI-generated content, a pattern the paper frames as an anthropomorphism paradox: features that increase engagement can also unsettle students precisely because they feel humanlike without being human. Interview subjects raised concerns that anthropomorphic AI could undermine originality, while others said it opened creative exploration. Across both strands of data, the paper reports a persistent theme that such tools can complement, but the authors state do not replace, direct interpersonal teaching.

Rights status

This is an education-perception study published under Frontiers in Education's open-access, peer-reviewed model; it makes no statement about copyright, licensing, or consent for AI-generated musical output, and none should be inferred from it. Its only findings concern how one sample of students perceived and used AI features in a learning setting.

What to check before you use it

This is an editorial checklist. The sample is a single population, Chinese undergraduate music students at multiple universities, gathered with self-reported, cross-sectional questionnaires the authors themselves flag as limiting causal claims; treat the findings as descriptive of that group rather than of music students, or the wider AI-music market, generally. Confirm whether a specific product or plugin family is named before implying the study endorses or condemns any commercial tool, since the study describes tested features generically. Note the qualitative interview subsample, 33 to 35 students, is smaller than the survey population and should not be quoted with survey-level statistical weight.

• Does a claim rest on the 600-student survey, the smaller interview sample, or both blended together?
• Is “students found it useful” being stretched into a claim about professional musicians or working producers?
• Does citing “AI cannot replace teachers” overstate what a self-reported, cross-sectional study can establish about causation?

The study's most durable point may be its paradox: the same anthropomorphic touches that make a practice tool feel responsive are the ones students found unsettling, a tension worth naming rather than resolving with a single feature update.