The intersection of artificial intelligence (AI) and music creation represents a complex, evolving dynamic carrying significant implications for human artistry. While AI undoubtedly augments certain technical aspects of the music production process, its ability to replicate the totality of the creative experience remains hotly debated. As this technological frontier rapidly advances, it is crucial to examine it through a measured lens — one that neither overstates AI’s capabilities nor dismisses its potential to fundamentally transform how music is created.
At its current stage, AI music creation tools operate by training machine learning models on vast datasets of existing musical compositions. This algorithmic approach allows AI to identify patterns and statistical tendencies which it can then utilise to generate new material like melodies, basslines, chord progressions and even complete songs within defined musical styles or genres. The likes of AIVA, Suno and Moises represent the vanguard of this AI composer toolkit.
However, this statistical repackaging of training data highlights AI’s core limitation: a lack of sentient contextual understanding. Music is inextricably linked to human experiences, emotions and the unique personal narratives its creators endeavour to convey. AI cannot inherently grasp the intimate stimuli and expressive intent behind why a particular musical motif or creative decision resonates or holds deeper meaning. Its output remains confined to regurgitating calculated probabilities based on the data it ingested.
‘It’s important to distinguish between assistive AI and generative AI,’ says Julia Rowan, Senior Policy and Public Affairs Manager at PRS for Music. ‘By assistive AI, we mean using AI as part of the composition process without necessarily substituting out the composer themselves. One use of AI that has existed for years is for mixing and mastering: it’s a way of augmenting the capacity of the tools that already exist.
‘Even generative AI tools can be used as a small part of the composition process that is otherwise carried out by a human. Music creators can use AI tools for inspiration and completely transform AI output, for want of a better term, with their creativity. I don’t think it’s quite as binary as not using AI or compromising one’s artistic integrity.’
This becomes even more apparent when examining the global impact of technological advances on music creation and consumption. The accessibility of recording software and hardware has opened Pandora’s box, empowering creators to inexpensively produce and digitally distribute their music worldwide. The rise of streaming platforms, social media and mobile technology has been similarly transformative. The likes of Bandcamp and SoundCloud have broken down geographical and economic barriers, allowing artists from across the world to share their music at the click of a button.
AI-driven data analytics tools help predict consumer trends and artist popularity, enabling record labels and marketers to better target their audiences. These tools analyse streaming data and social media engagement to provide insights into what music is trending, which is invaluable in an industry as dynamic as music.
‘It’s important to distinguish between assistive AI and generative AI in music.’ - Julia Rowan
In this emerging AI-assisted age, technologists and composers may be better served by forming a symbiotic relationship akin to an extended creative collaboration. AI could help identify unconscious creative tendencies, inspiring new modes of expression by revealing novel possibilities beyond conventional paradigms. It could augment a musician’s toolkit while still preserving their autonomy as the prime creator and curator.
Organisations like PRS for Music are playing a pivotal role in stewarding the responsible development of AI in music. Back in February PRS published its AI principles, a proposed balanced framework for how AI should be ethically used in music creation while promoting and protecting human creativity.
PRS’s AI principles establish a balanced stance: avoiding outright rejection of AI while implementing pragmatic governance to mitigate risks like plagiarised outputs, diminished human agency or creator exploitation. The focus remains squarely on empowering and protecting PRS’s human composer members as they judiciously determine if and how to incorporate AI’s emerging creative capacities. These principles aim to provide initial guidance, mandating transparency through clear attribution when AI is involved, while still reserving full royalty payments and membership benefits for human creators maintaining oversight of the process.
Asked whether there needs to be a change in copyright law to keep pace with AI technology and protect creators, Julia Rowan says: ‘The UK copyright regime is one of the best in the world, we don’t need a wholesale change of that framework. Instead, it’s about putting some AI-related safeguards in place to ensure the copyright regime remains de facto valid and the core principles of authorisation and remuneration still stand. What I mean by that is, for instance, clarification that authorisation ought to be sought and obtained by AI developers if they want to train their models on copyright-protected works. As a creator you should also be entitled to compensation for that use, which is the second important part of this picture.
‘What’s missing in legislation is a transparency regime for AI,’ she continues. ‘We need regulation that would place an obligation on AI developers to disclose what their models have been trained on in a way that enables enforcement of their rights by music creators. Without record-keeping of training data, music creators won’t be able to know whether their works are being used by AI models.
‘Finally, a big concern that is emerging around the world and cross-sector, not just in the creative industries, is around provenance and authenticity of content. People ought to be able to know whether the content they are consuming was generated by AI or by human beings. This is not only a matter of consumer choice and protection, but it even relates to the integrity of democratic processes when it comes to deepfakes.’
Ultimately, AI music creation tools could serve as viable supplementary tools to augment and stimulate the artist’s personal creative expression, not automate it entirely — but only if robust safeguards are implemented. Their works must be protected from uncompensated use as training data, with copyrights and royalties clearly upheld through an established licensing framework. With these crucial protections in place, AI’s maturation represents an evolution that musicians can help consciously shape through proactive exploration, experimentation and open cross-disciplinary dialogue, rather than a forced obsolescence of human artistry.
This article features in a special edition of M Magazine celebrating 110 years of PRS. You can read the magazine here.
