‘AI cannot answer the most important questions’: Benoît Granier on music education, creative identity and Sonic Futures 2026

When Benoît Granier stood at the podium of Xi’an Conservatory of Music on 12 June 2026, it was more than a professional engagement. It was his first time back in China since Covid, returning to a network of relationships built over more than a decade with institutions including the Central Conservatory of Music, Beijing University and collaborators across Beijing and beyond.

He had been invited to open Sonic Futures 2026 – 声之未来：音乐·媒体·人工智能 – an international conference exploring music, media and artificial intelligence. Alongside him in the opening ceremony were Dr Fan Fan, director of Huawei’s Central Media Technology acoustics laboratory, and Kees van Oostrum, former president of the American Society of Cinematographers. Behind them sat conservatoire faculty from across China, researchers, representatives from Huawei and BYD’s automotive technology division, and students trying to make sense of where music, media and AI might be heading next.

‘This was not an academic gathering debating whether AI belongs in music,’ Granier reflects. ‘This was a room where music, technology and industrial scale had already decided to sit down together. The question they were asking was not should we? It was how?’

For Granier, who is Course Director for Music and Audio Production at Coventry University’s School of Art and Creative Industries, as well as a trustee of Coventry Music Trust and the Charles Barratt Foundation and CEO of Artful Technology, that distinction matters. His interest is not in AI as a shortcut to creative output, but in what it can reveal about the conditions in which students create, the frustrations that stop them and the tools they might need to build for themselves.

‘I didn’t come to Xi’an to report on what others are doing with AI,’ he says. ‘I came to show what I had built – and to make an argument about what that means for how we teach, and how we create.’

The barline problem

At the centre of Granier’s argument is a deceptively simple problem: the barline.

‘When you write music using notation software, the software makes assumptions,’ he explains. ‘It assumes metre. It assumes the bar. It organises your musical thinking around a grid that was designed for a certain kind of music – and works against you the moment you move outside it.’

For many composers, that grid is invisible until it becomes an obstacle. But for music that does not sit comfortably within regular metre – Granier cites Messiaen as an example – standard notation software can force a creative compromise.

‘You face a choice: fight the tool, or simplify the music. Most people, under time pressure, simplify the music. And in doing so, they simplify their voice.’

Granier’s response was not to use AI to generate music, but to use it as an assistant in building the tools he needed. These were tools designed around his own creative logic: his sense of rhythm, sonic friction and musical feel.

‘I refused that choice,’ he says. ‘Instead, I used AI as an assistant to build the tools I actually needed – tools designed around my own creative logic, my own sonic frustrations, my own sense of what the music should feel like.’

This, he argues, is where the real educational opportunity lies.

‘AI is most powerful not when it creates for you, but when it helps you build the conditions in which you can create more fully.’

AI as assistant, not author

That distinction – AI as assistant rather than author – was central to Granier’s presentation in Xi’an.

‘I did not use AI to generate my music,’ he says. ‘I used AI to remove the obstacles between my imagination and its realisation. The result is music that could only have come from me, but made more possible because the tools were made for me.’

For music educators, the implications are significant. Granier believes that much current discussion around AI risks training students to become consumers of existing tools rather than authors of their own creative processes.

‘We point them toward existing platforms, existing workflows, existing assumptions about what AI-assisted music looks and sounds like,’ he says. ‘And in doing so, we risk doing exactly what the notation software does: fitting the student’s creative voice to the tool’s logic, rather than the other way around.’

For Granier, the most important shift is to reverse that relationship.

‘The invitation to flip that relationship is the most important pedagogical opportunity AI has given us.’

Across the two days of Sonic Futures, Granier took part in the opening ceremony, joined an international roundtable and delivered the session ‘From Patch to Prompt: Building AI-Assisted Tools for the Creative Musician’. What struck him most was not technological spectacle, but the seriousness of the educational conversation.

‘The roundtable surprised me, not because of disagreement, but because of recognition,’ he says. ‘Professors and artists who might have been expected to defend traditional practice were instead asking how. How do we integrate AI without losing what matters? How do we teach students to use these tools without teaching them to depend on them? How do we ensure that technology enhances the creative voice rather than replacing it?’

For Granier, these were the right questions – and questions that music education internationally has been slow to ask with sufficient clarity.

‘Students were taking notes throughout. That, more than any formal response, told me the conversation had landed somewhere real.’

AI in the classroom

Back in Coventry, Granier has already been testing these ideas in his own teaching.

Last year, his students worked on an AI band project, creating a piece of music using AI tools alone. The purpose was not to celebrate the technology’s output, but to expose its limits.

‘The limitations were as instructive as the outputs,’ he says.

More recently, he has begun using AI not to generate creative work, but to help students understand and articulate their creative identity. In one exercise, students ask AI to develop a colour palette for their portfolio website, based on their stated mission, vision and body of work.

The point is not the colour palette itself. It is the briefing process.

‘Students have to know themselves well enough to brief the machine meaningfully,’ Granier explains. ‘That is where the real pedagogical shift lies – not in what AI can make, but in what it reveals about what students already know, and don’t yet know, about themselves.’

He describes this as ‘AI as mirror, not generator’.

It is a phrase that captures much of his approach. The value of AI, in this model, is not that it replaces artistic judgement, but that it forces students to confront the clarity or otherwise of their own intentions. A weak prompt can expose a weak idea. A vague brief can expose a vague artistic identity. The machine does not solve that problem. It reveals it.

The questions AI cannot answer

In Xi’an, Granier ended his day two session not with conclusions, but with questions for the students in the room.

‘I asked them to carry three questions into their practice,’ he says.

The first was: ‘What frustrates you?’

‘Not “what could be improved”,’ he explains, ‘but what actually stops you? What do you avoid because it costs too much time or skill?’

The second was: ‘What would it need to do?’

‘Describe it to someone who cannot see inside your head. What is the simplest version that proves the idea works?’

The third was: ‘What would it feel like?’

‘Not how it works technically. How does it feel? What is the aesthetic experience of using it?’

For Granier, these are not technical questions. They are artistic questions. And precisely because they are artistic questions, they remain beyond the reach of the machine.

‘These are the questions AI cannot answer,’ he says. ‘Which means they are the most important questions we have.’

He was later told that he was the only speaker at the conference to end with questions for students rather than answers. That feels, in many ways, like the point. In an environment where AI is often framed in terms of solutions, Granier’s emphasis is on the continued necessity of human intention, judgement and meaning.

Keeping the artist in the process

For Granier, the anxiety surrounding AI in music is not really about technology. It is about value.

‘It is about whether artistic practice still has meaning in a world where machines can generate, imitate and iterate at scale,’ he says.

His answer is not a nostalgic defence of human creativity against technological change. It is a more demanding proposition: that human creative responsibility becomes more important, not less, when tools become more powerful.

‘The machine can produce. It cannot mean. That remains entirely our responsibility.’

That idea has direct implications for music education. If students are to work meaningfully with AI, they need more than technical fluency. They need a stronger sense of who they are creatively, what frustrates them, what they want to make possible and what they want their work to feel like.

‘The future of music education is not about teaching students which AI tools to use,’ Granier says. ‘It is about making them so clear about their own creative voice – so honest about their frustrations, so precise about what they need, so alive to what they want things to feel like – that they can direct any tool, including AI, toward something genuinely their own.’

That was the message he took to Xi’an. Not that AI should replace creative practice, but that it should challenge educators to teach creative identity more seriously.

‘That is what I came to Xi’an to say,’ he says. ‘And building the tools was how I proved it.’

Benoît Granier is Course Director for Music and Audio Production at Coventry University’s School of Art and Creative Industries, a trustee of Coventry Music Trust and the Charles Barratt Foundation, and CEO of Artful Technology. He delivered the opening address, participated in the international roundtable and premiered ‘Cubiculum Soni / 山水’ at Sonic Futures 2026, Xi’an Conservatory of Music, June 2026.