Professor Pat Thomson, CLA’s Senior Evidence Associate, reports on a research paper examining the impact of AI on school music education, asking key questions about equity, ethics and artistic expression, and making important recommendations, including for future research.
Cheng, Lee (2025) The impact of generative AI on school music education: Challenges and recommendations, Arts Education Policy Review, 126:4, 255-262, DOI: 10.1080/10632913.2025.2451373
Generative AI (Gen AI) is a prime example of change happening ahead of the production of research evidence. But there are people who are thinking deeply about the topic and pulling together the research that is emerging.
Lee Cheng’s paper tackles a question that might keep arts educators up at night: how do we respond to generative AI that can compose music, create images, and produce creative content with just a few typed prompts? This paper focuses on music education but speaks to challenges facing all arts disciplines as Gen AI tools become ubiquitous in student lives and across the creative industries.
The paper presents an optimistic case for Gen AI in arts education. Lee Cheng argues that Gen AI breaks down traditional barriers to creative production. Students who lack years of technical training can suddenly generate sophisticated musical compositions, visual artwork, or even choreographic ideas. Research is cited to show that students using AI-assisted learning achieve better outcomes and feel more motivated to engage with creative work.
There is clearly some potential with Gen AI to democratise artistic expression and address long-standing equity issues where expensive instruments, materials and/or specialised training have favoured privileged students. AI can also enhance theoretical learning—imagine students having conversations with simulated historical artists or analysing how different creative styles emerge from various cultural contexts.
But the concerns are equally substantial and can’t be easily dismissed. Cultural bias sits at the heart of the problem, Cheng says. Most AI training datasets heavily favour Western art forms and perspectives, which means generated content tends to reproduce and amplify these art forms, traditions and genres rather than celebrating the diversity arts education aims to cultivate. Whether we’re talking about music composition, visual design, or dramatic interpretation, AI systems trained predominantly on Western sources risk eroding the multicultural dimensions espoused in contemporary arts education.
Then there’s the creativity question that gets to the core of arts teaching. Yes, AI enables production, but what kind of creative development happens when students generate artwork with minimal intellectual or emotional engagement? While some celebrate “co-creation,” there’s a concern that overreliance on AI becomes a crutch that undermines creative practice. Students might stop pushing beyond algorithmic suggestions, losing the struggle and experimentation that builds genuine artistic capacity. Gen AI can generate unpredictable results that function as “black boxes” where neither teachers nor students really understand how decisions were made.
Equity issues also multiply in unexpected ways. Premium Gen AI features sit behind paywalls, instantly creating classroom haves and have-nots. Cheng warns that some students face disproportionate accusations of cheating when using AI tools, perpetuating existing inequities. And without ethical frameworks, students might misuse gen AI in harmful ways, from generating inappropriate content to creating deepfakes targeting classmates.
So what should arts educators do? Cheng’s recommendations translate across disciplines. First, developing AI literacy is non-negotiable. Arts teachers need domain-specific understanding of how AI interprets visual composition, musical structure, movement, or dramatic narrative. Students should learn to recognise when AI reproduces biased perspectives, understand the difference between human and machine creativities, and explore applications beyond simple generation. Cheng is especially concerned about the need for radical rethinking of assessment across all arts subjects. Current rubrics may not distinguish between student creativity and AI output. Rather than fixating on polished final products (which AI excels at producing), assessment could emphasise creative process and reflection. Cheng suggests strategies such as asking students to document their prompts, describing iterations and dead ends, explaining artistic choices, and reflecting on what AI contributed versus their own vision.
Second, access issues need immediate attention. Schools and arts teachers should advocate for funding models similar to those supporting technology acquisition, ensuring all students can engage with the tools that are already shaping creative industries. When resources fall short, Cheng suggests, supervised access during class time beats outright bans that students will circumvent while learning nothing about responsible use.
Third, professional development, particularly for arts educators who may feel less confident with emerging technologies, is vital. Training should not just cover technical skills but should address pedagogical strategies for maintaining artistic integrity while embracing new tools, understanding copyright implications specific to creative work, and recognising how AI impacts diversity and equity in student learning.
Finally, Cheng also identifies a number of research needs, in particular, longitudinal studies which track how AI-assisted creativities develops across all arts disciplines over time. How do students who learn with AI tools perform later? What assessment models actually work in practice? How do different student populations experience these tools, and what pedagogical approaches best support underrepresented groups? How can we address cultural bias in ways that honour diverse artistic traditions?
Although writing about music, Cheng’s ultimate argument applies broadly: arts education must engage with Gen AI to remain relevant while establishing thoughtful policies protecting what matters most, artistic development, cultural diversity, equitable access, and the human dimensions of artistic expression. This isn’t about choosing between tradition and technology Cheng says, but about shaping how these powerful tools can serve rather than undermine arts education’s core purposes.
