Is AI killing music

If you want to know how Artificial Intelligence (AI) is affecting the creative industries, a routine Google search can make for pretty grim reading: graphic artists fear for their jobs and musicians are worried about their sound being pilfered, while in Hollywood, writers have already been out on strike - demanding protection from the creeping use of AI-generated scripts.

But are these fears justified? In this Big Question, composer, performer, and sound artist Federico Reuben, and CoSTAR Live Lab Co-Director Gavin Kearney, tell us how they think AI could cross the legal badlands and emerge from the shadows, to become the next chapter in creativity; and how institutions like the University of York could play its part in creating a new perception of technology’s latest bogeyman…

Myriad musical backgrounds

Federico Reuben is a musician and academic with an enviable CV: a gifted pianist from an early age, he played in a number of styles in his native Costa Rica, with stints in classical, jazz, pop and reggae musical outfits all influencing his musical outlook.

His musical studies took him to the US and then to the prestigious Institute of Sonology in the Netherlands’ Royal Conservatoire, before embarking on his PhD in London’s Brunel University. He’s been in the UK ever since, and now is a Senior Lecturer in the School of Arts and Creative Technologies.

But there’s an unusual string to Federico’s bow. Far from feeling threatened by artificial intelligence, he sees AI as a rich and exciting creative tool. Last year his free jazz trio, Sveið, released its album Latent Imprints, on which he performed as the ‘live coder’, enabling drummer Emil Karlsen and saxophonist James Mainwaring to improvise with spontaneously-produced AI sounds, and the band regularly reproduces this format on stage.

Federico explains that, far from ‘killing’ music, AI could - with the right legislation in place - point the way to an entirely new musical landscape.

“To understand whether or not AI could kill music, you first need to look at what AI really is," he says. “For me, AI is about machine learning, and the ability that algorithms have to learn from data.

“My first encounters with AI were when I started to use algorithms to write and perform music through my exposure to computer programming.

“But for a lot of people, when they think about AI and music, they’re thinking about these chatbot systems, where you have a box for a line of text, you type in what you want and music comes out; at the moment, we’re fixated with these kinds of ‘text-to-something’ systems.”

A portrait photo of Federico in front of some industrial machinery — “I started to use algorithms to write and perform music through my exposure to computer programming.”

“I started to use algorithms to write and perform music through my exposure to computer programming.”

Photography by Michael Hodges

Code or conduct

But, says Federico, such systems really only scratch the surface of what AI is, and what it can do. He’s been interested in ‘live coding’ since his student days, a seemingly dizzying method of writing computer code on the spot during live performance.

“For me it’s not so different from playing the piano, it has elements of improvisation and elements of composition but it happens on the spot and there are elements of performance as well.”

But he admits there are differences too. The physicality, so often crucial in musical performance and a trademark of performers like Charlie Mingus, John Coltrane and Ella Fitzgerald, isn’t the same.

“If you play a double bass you’re really getting your fingers dirty,” says Federico. “You’re applying real force and it’s a very distinct type of embodied performance which comes across in the sound. Live coding is quite cerebral. It uses mental processes that have to do less with bodily interaction and more to do with using your ears and certain parts of your brain.

“Live coding is quite cerebral. It uses mental processes that have to do less with bodily interaction and more to do with using your ears and certain parts of your brain.”

“That’s a difference that’s really interesting. Of course all music making requires a combination of embodied and high-level cognitive functions, but in live coding as well as in some conventional instrument playing, like playing Bach on the piano, there’s less focus on physical engagement with the instrument and more with slow, deliberate, and analytical mental processes. So the fact that I started out as a pianist and have developed into live coding makes a lot of sense.”

Mistrusting the machines

Federico also believes the ‘text-to-music’ AI systems grab the headlines because of a general mistrust of the technology, but he emphasises that, in some ways, this type of music making isn’t all artificial and removed from human creativity.

“People tend to think of AI as being this robot-like disembodied entity that comes up with something which only a human should be able to create. But in reality, the outputs these models generate do contain a lot of human input. The algorithms are trained on human experience - if you hear a voice, there’s been a human there before, as well as musicians, engineers, computer scientists; there’s a lot of knowledge and experience that is human.

“The issues that are problematic tend to be around people’s identity. With this sort of AI you can take Thom Yorke’s voice and make him sing, or say, things he’s never uttered before. The voice is a very personal thing so it’s reasonable to be scared of that.

“The same goes for recorded music. If you spend years working on a body of music, someone can train an algorithm in that music and create something very similar to your musical identity.

“This sort of technology can be misused and I think that’s the problem people have. You can’t always trust that the person singing is real any more.

“But when the phonograph reproduced a disembodied human voice for the first time in the late nineteenth century, people were scared too, and when music was first recorded, people thought musicians were going to lose their jobs. AI is a similar paradigm shift, another disruptive technology.”

Code or conduct

But, says Federico, such systems really only scratch the surface of what AI is, and what it can do. He’s been interested in ‘live coding’ since his student days, a seemingly dizzying method of writing computer code on the spot during live performance.

“For me it’s not so different from playing the piano, it has elements of improvisation and elements of composition but it happens on the spot and there are elements of performance as well.”

But he admits there are differences too. The physicality, so often crucial in musical performance and a trademark of performers like Charlie Mingus, John Coltrane and Ella Fitzgerald, isn’t the same.

“If you play a double bass you’re really getting your fingers dirty,” says Federico. “You’re applying real force and it’s a very distinct type of embodied performance which comes across in the sound. Live coding is quite cerebral. It uses mental processes that have to do less with bodily interaction and more to do with using your ears and certain parts of your brain.

“Live coding is quite cerebral. It uses mental processes that have to do less with bodily interaction and more to do with using your ears and certain parts of your brain.”

“That’s a difference that’s really interesting. Of course all music making requires a combination of embodied and high-level cognitive functions, but in live coding as well as in some conventional instrument playing, like playing Bach on the piano, there’s less focus on physical engagement with the instrument and more with slow, deliberate, and analytical mental processes. So the fact that I started out as a pianist and have developed into live coding makes a lot of sense.”

Mistrusting the machines

Federico also believes the ‘text-to-music’ AI systems grab the headlines because of a general mistrust of the technology, but he emphasises that, in some ways, this type of music making isn’t all artificial and removed from human creativity.

“People tend to think of AI as being this robot-like disembodied entity that comes up with something which only a human should be able to create. But in reality, the outputs these models generate do contain a lot of human input. The algorithms are trained on human experience - if you hear a voice, there’s been a human there before, as well as musicians, engineers, computer scientists; there’s a lot of knowledge and experience that is human.

“The issues that are problematic tend to be around people’s identity. With this sort of AI you can take Thom Yorke’s voice and make him sing, or say, things he’s never uttered before. The voice is a very personal thing so it’s reasonable to be scared of that.

“The same goes for recorded music. If you spend years working on a body of music, someone can train an algorithm in that music and create something very similar to your musical identity.

“This sort of technology can be misused and I think that’s the problem people have. You can’t always trust that the person singing is real any more.

“But when the phonograph reproduced a disembodied human voice for the first time in the late nineteenth century, people were scared too, and when music was first recorded, people thought musicians were going to lose their jobs. AI is a similar paradigm shift, another disruptive technology.”

The legal imperative

And for Federico, this is where the subject of legislation takes centre-stage. A working musician himself, he knows the importance of a coherent, enforceable legal system to ensure musicians get paid.

“The copyright issue is complicated because it’s about attribution,” he says. “The issue here is that people with specialist knowledge, resources and power are taking everything that’s on the internet and using it to train AI models without proper copyright attribution.”

He says copyright law was already struggling to effectively deal with music streaming services, and, in its present form, it simply isn’t able to cope with AI.

“But this is more of an economic problem than it is a creative problem. I think Noam Chomsky referred to it as ‘high-tech plagiarism’, and there’s a feeling that with regard to the creative arts, AI could be a copyright-avoiding machine. There are a lot of interesting questions about this, and I think we’re going to have to find creative answers to them.”

Drawing comfort from history

But if those answers can be found and the legalities can be upheld, then there’s every chance that musicians of the future could see AI as an ally in the now vast array of technology inherent in the musical process - most of which was viewed with suspicion when first launched.

“At the moment the narrative with music AI is around AI-generated music getting into the charts, or the idea of AI being a robotic entity which is ‘replacing’ musicians.

“I can see a comparison with sampling, which in its early days was seen as a threat both to musicianship and to copyright. It took time for musicians to realise that sampling is a new instrument and to see the possibilities it presented.

“So I hope something like that will happen with AI. That people will start to see it as a co-creating relationship, where musicians and AI are jamming and making music together, creating new sensations, emotions and knowledge. That’s when people will start getting less fearful and more excited about it.”

A crowd of people with their backs to the camera, looking towards a stage dressed with political campaign flags in the background. On the stage stands a group of figures who are leading the political rally at a location in India. — “There are a lot of interesting questions… we’re going to have to find creative answers to them.”

“There are a lot of interesting questions… we’re going to have to find creative answers to them.”

Photography by Kazimierz Ździebło

Photography by Alex Holland

Photography by Dave Walker

Creative connections

And there are other creative industries to consider too. Over in West Yorkshire, a new research and development facility is set to become one of the UK’s foremost creative centres. Launched in February 2025, CoSTAR Live Lab is redefining possibilities in storytelling and the audience experience, be it in film, theatre and the live music arena.

Gavin Kearney is one of CoSTAR’s Directors and from his vantage point, he says he can see artists and creators are beginning to see the creative possibilities that AI can hold.

“We're seeing strong evidence of AI being used to augment rather than replace creative practice. Both musicians and technologists are using it to support ideation, accelerate production workflows, and enable new forms of interaction between performers, audiences and environments.

“In live performance in particular, AI opens up possibilities for adaptive and responsive experiences where sound, visuals or narratives can evolve in real-time. It also lowers technical barriers, allowing smaller teams or independent creators to realise ideas that previously required significant infrastructure.

“The most interesting work isn’t about generating finished content, but about enabling new creative processes and forms of collaboration between humans and machines.”

Creating Spaces

Like Federico though, Gavin agrees that anyone making a living in the creative industry is entitled to feel nervous about some technological developments, particularly when the speed of those developments seem to outrun the legal changes being put in place to protect artists.

But here, he says organisations like CoSTAR Live Lab and the University of York, can help.

“Some concerns are absolutely valid, particularly around authorship, attribution and economic impact. As Federico has said, if artists’ work is used to train systems without consent or compensation, that undermines the sustainability of the sector. There are also legitimate concerns around identity, especially with voice and likeness.

"The role of research institutions is to engage with these issues directly. That means developing transparent and ethical approaches to data use, contributing to standards and policy discussions, and working closely with artists to ensure new tools reflect their needs and values.

“It’s also about creating spaces like CoSTAR Live Lab where creatives can experiment with AI in a supported way, so they can understand its capabilities and limitations, rather than being positioned as passive recipients of change.”

And Gavin admits that, when the subject of AI comes up, he does encounter degrees of trepidation, but also a high degree of curiosity, which is where research, teaching and learning is playing a crucial role.

“There is definitely concern, particularly from early-career creatives and freelancers who feel more exposed to disruption. But alongside that, there is also curiosity and genuine excitement.

“When people have the opportunity to engage with these tools directly, the conversation often shifts. It becomes less about replacement and more about how AI can be incorporated into practice in a way that remains distinctly human.

“What we’re seeing is that fear tends to come from distance, whereas hands-on engagement tends to lead to more nuanced and constructive perspectives.”

Photography by Alex Holland