What can creatives bring to AI?
The creative industries need to think bigger when it comes to AI
I’ve been thinking a lot about classical music and AI this month. As an industry broadly, classical music hasn’t been hugely impacted by AI yet. Pioneers in music and AI have been working with various kinds of artificial intelligence for nearly fifty years, but it’s remained a relatively niche area of the industry. It’s only in the last couple of years that the rapid advances in frontier AI — the most advanced machine learning models available, like Claude or ChatGPT — have put AI firmly on the radars and agendas of most working musicians.
So the Royal Ballet and Opera ran a four-day festival, SHIFT, to explore two questions: “What can AI offer creatives? And what can creatives offer the world in the age of AI?” These feel like two of the most pressing questions facing creatives at the moment, and the festival brought together practitioners across the world to showcase work using different forms of AI, and explore ethical issues around AI broadly. I reviewed it for Opera Now (read here), and interviewed some of the artists involved for an earlier Financial Times piece about the way AI is impacting on classical music (read here).
I loved the sense of exploration and innovation at SHIFT. There were so many positive examples of artists using AI ethically and responsibly. Paola Prestini’s opera Sensorium Ex was a standout, giving minimally verbal individuals a voice on the opera stage through AI systems that expand vocal expressivity. Yet it always felt like there was an elephant in the room that wasn’t being addressed, and that elephant was frontier AI itself. SHIFT felt like an oasis — or, perhaps, a cocoon in which the arts could hide from the rest of the world rather than engage directly with it. This was how I summed it up for Opera Now:
Musicians engaging with AI are not working in a vacuum. Although they might have laudable, even utopian visions for what AI may in theory be able to bring the world, there is a fundamental disconnect between the material resources and epistemic priorities of these artists, and the Big Tech companies who are driving AI development. Speaking after Sensorium Ex, disability activist Jacob Nossell said that ‘to have a voice is everything’ – but it also matters who is listening to those voices. Given the gulf between the uses of AI on display inside and outside the opera house, it doesn’t seem like Big Tech is interested in listening to artists.
Local examples of using Midjourney to generate costume designs are all well and good, but some researchers forecast that we are about four years away from superintelligent AI that is capable of outperforming human researchers in a variety of environments. There is disagreement over the timeline of these forecasts, but it’s by a matter of years, not centuries, given how quickly AI’s coding capabilities are improving. It’s plausible that on a five to ten year horizon frontier AI completely changes the way we work, think, share information, and interact globally. If we aren’t addressing this as an industry, I don’t think we’re really addressing what AI might do for creatives, or vice versa. We aren’t thinking existentially enough.
Perhaps, though, SHIFT inadvertently revealed the single most important thing the creative arts can offer in an AI-saturated age, and that is to pull us back to the human. Over four days of work interacting with AI, I was most gripped by the moments when the human was foregrounded — the sheer audacity of dancers stretching their bodies to the bounds of the physically possible, the quavering imperfection of a human voice, the pleasure of an unexpected conversation with the person next to me.
Human and machine don’t have to be in conflict, and there are without question benefits that some forms of AI can bring humanity. But creative work — and the very physical forms of ballet and opera especially — can powerfully demonstrate what we stand to lose from unchecked, rapid automation. Even if Big Tech is not listening to artists, audiences are, and public opinion about the way technologies are being developed and deployed matters. Tech companies are not completely beyond governance yet. Just today, the UK has followed Australia in banning social media for under 16s. This will disproportionately impact on Meta, who aren’t major players when it comes to AI, but the principle stands that humans do have a say over the amount of power we are prepared to give to a small handful of companies. A reminder that humans still have a collective voice, still have agency, might turn out to be the single most important thing that creatives can offer the world in the age of AI.
You can read my review of SHIFT for Opera Now here, and my article about the way AI is changing classical music in the Financial Times here.

I agree. It's worth differentiating though between frontier AI (e.g. ChatGPT, often trained using copyright work and with huge environmental costs), and the local models that the artists at SHIFT were predominantly using, which all used consensual data and don't have the same kind of environmental impact. It feels like there there are useful versions of AI that are small-scale, local, and are designed as human tools - but that's not what frontier AI is.
For me, one of the biggest issues with A.I. is the fact that it is built on stolen art, music, literature, etc. For that reason alone I feel artists of all kinds should not use it. And don’t forget the huge environmental cost. Our planet is already is crisis, we don’t need anymore huge consumers.
We don’t NEED A.I. period.