Is AI really the future of music?

Clips of animated cats having extramarital affairs as you doomscroll through Instagram reels, uncanny figures replacing models in the ads plastered on tube platforms, maybe even an overly-therapised message from a friend after a disagreement. Whether you know it or not, AI has now transcended bad sci-fi movie plots and the pipe dreams of tech bros and seeped into daily life, becoming as ubiquitous as the microplastics clogging our veins. Now, its eerie presence on our playlists is even threatening to shift the way we interact with music – an art form as old as humanity itself.

Earlier this month, Spotify revealed it had removed over 75 million “spammy” AI-generated songs from the platform, while Deezer recently claimed that over 28 per cent of tracks uploaded are fully created by AI – a drastic increase from the 10 per cent they reported in January. Some of these fake artists have managed to form hefty fanbases, too. Psych-rock “band” The Velvet Sundown racked up millions of listeners on Spotify before reports emerged that they were merely the product of AI software. “It’s rare for a band to just materialise with a debut album,” their (most likely Chat GPT-penned) artist bio ominously reads. 

Cara* was among those duped by the band, and describes feeling “absolutely flabbergasted” upon discovering that they didn’t exist in the real world. “Finding out it wasn’t ‘real’ human-produced music made me incredibly uncomfortable and freaked out to the point I didn’t want to listen to it anymore,” she says, adding that she stopped listening “almost in protest” against the algorithm promoting the band any further.

Last week, though, the algorithm still fed her another AI-generated artist – this time a hard rock band called Bleeding Verse. “The song I heard was really perfect timing for the emo mood I was in,” she says, recalling texting her brother, who’d also come across their music that day, to bond over “how legit sounding this new band was.” When she got home, though, a quick Google search revealed the band to be yet another AI catfish. “My heart sank and my mind melted,” she says. “It’s impossible to comprehend how we’ve gotten here; it floored me all week.”

Over the past century, technology has shifted the way we experience music, only to be met with controversy and outrage at each step. However, the introduction of AI comes at a particularly precarious era for artists, with streamers forging a culture of passive listening and background “muzak”. If users aren’t tapping “skip” when AI-generated tracks are pushed to them via the algorithm, streamers may only be rubbing their hands together, caring less about the destruction of art and more that they get to pay these faux artists even less of a fraction of a penny than the real ones. 

Some “artists” are even getting major industry attention. Xania Monet, an AI-powered R&B singer, was reportedly at the centre of a bidding war between multiple labels earlier this year, with one apparently offering a $3 million record deal. Monet, who appears in short clips promoting her songs while washed in the telltale yellowish AI sheen, was created by 31-year-old, Mississippi-based poet Talisha Jones, who clicked Monet into existence using Suno – the software company already at the centre of multiple copyright lawsuits.

News of fake artists signing deals that would be life-changing for real musicians can seem like the beginning of the end for the music industry. Tom Collins, professor at the Department of Music Engineering Technology, Frost School of Music, University of Miami, however, theorises that cases such as Monet are likely just “publicity stunts” from labels, while the actual popularity of AI-generated artists is also hard to truly decipher. “You have a human uploading AI-generated music, which is then commented on by AI bots. Human content consumers, journalists, and so on, are now trying to provide commentary on bot-bot interactions,” he explains. “I expect some similar things are happening on Spotify: a human uploads AI-generated music, and I expect some of the streams are bot-related.”

I don’t know what is real or what is fake anymore. I want to go back so, so badly

It’s encouraging to hear that AI musicians likely aren’t really in danger of replacing real ones. But even if they are just a strange marketing tactic, while AI creations like Monet garner industry attention, the artists that her vocals were trained on currently receive no compensation for the use of their copyrighted work. In the UK, the government has even proposed law changes that would allow AI developers to train their models on copyrighted work, without the original creators’ consent or payment.

The plans have been criticised by musicians like Paul McCartney, Dua Lipa and Elton John, but the fact that copyright, which was signed into British law in 1710, is even up for debate has rung alarm bells across the industry. “It’s a simple and important right and philosophy, one we have taken for granted, that is now under threat,” Sarah Pearson, CEO of Beyond The Music – an annual conference that addresses challenges in the music industry – explains. “This is a vital global cultural moment; we must be on the right side of history and protect the precious rights of the creator.”

At this tipping point, labels will need to come to an agreement with AI companies. Already, Universal Music and Warner Music are reportedly nearing landmark deals with tech groups. Yet the music industry has notoriously and historically undercut artists with each technological development, so fear that further exploitation is ahead is rife, especially for independent artists without representation.

For fans, knowing that music – something so intrinsic to our identity as human beings – can be chewed up and spat out in the most listenable form by an AI bot is also another depressing symptom of our dystopian times. “I don’t know what is real or what is fake anymore,” Cara laments. “I want to go back so, so badly. I’m just like ‘stop the train, stop the music, I want to get the F off!’”

To prevent a future where beige, safe music void of innovation or human emotion is the norm, some industry leaders are laying out regulations of their own. Deezer have already adopted a full transparency approach to AI-generated music, launching an AI detection tool and tagging music that is 100 per cent AI-generated, as well as excluding these songs from algorithmic and editorial recommendations – meaning they won’t be prioritised or promoted over the work of real artists.

Similarly, Spotify recently unveiled a tool that will allow artists to declare if their work was made with AI, though at present this is only on a voluntary basis. Though their action speaks to a wider rejection of fully AI-generated music from consumers like Cara, coming from the biggest streamer, the move is deeply inadequate and seems more of a PR spin amid controversy surrounding current CEO Daniel Ek (who will step down in January to become executive chairman) investing in AI military drones. 

This isn’t to say there is no room for AI to enhance creativity when used responsibly. “We need time to harness the advantages,” says Beyond The Music’s Oli Wilson. The popularity of a handful of fake artists also doesn’t mean our natural inclination towards the great things about music – connection, culture and community – will cease. “Commercially, sure, people will listen to AI music, but I believe people will always want human music too, and there will always be a market for that,” Wilson says.

As consumers, we must resist the tidal wave of mundane, safe AI-generated music and support real artists. After all, as Collins puts it: “[Music] is about a human expressing something that is then perceived by another human. Even if there is a huge amount of technology in between the humans at either end of this process, music is ultimately a form of human-human communication.”