AI makes ‘a scary time’ for musicians
While Hamilton heavy-metaller Paul Martin has spent decades creating music with a distinct sound like heavy machinery, it’s all come from the heart.
But even he now wonders if the robots have finally come for him.
Fresh from flooding the internet with artificially generated Leonardo da Vinci-level paintings last year, and writing essays for crafty students, AI is now ready to battle humans over music.
‘‘It’s scary as a songwriter thinking, wow, there’s this technology that could just make you 100% redundant,’’ muses Martin . . . and he’s not alone in the industry.
Martin, something of a godfather in the Kiwi metal scene, has brought metal to the masses since the 90s, first through his Axe Attack radio show and bands including World War Four and his latest – Devilskin, which has toured offshore and gained commercial success.
While technology has been a part of music for over a century now, professionals reckon fullyfledged songs can now be produced with minimal input.
Martin, who wields bass for Devilskin, says in the future it will be hard to distinguish between music created by humans and robots – though it won’t have the ‘‘beautiful connection’’ between musicians and audience.
At the beginning of this year, Google released an academic
paper discussing its MusicLM generative AI model that makes music from users’ text prompts.
The model can make a new song or take an existing song and produce it with a different sound.
Martin said Devilskin was up against AI, an ‘‘unseen allknowing monster’’, and the only way they could set themselves apart was to be themselves.
‘‘It’s crazy because we get people contact us and say how certain songs have changed their lives and moved them in beautiful ways. Our music gets played at
funerals and weddings, people get the lyrics tattooed on the arm.
‘‘So when you’re touching people that closely, I think it’s just this beautiful connection that I couldn’t imagine doing it artificially.’’
University of Auckland senior lecturer and co-ordinator of music technology Dr Fabio Morreale has done extensive research on the impact of technology on music.
AI could be a great tool for composers and musicians, or extremely dangerous, Morreale said.
‘‘Or, at least, let’s say that it risks redefining completely the way people listen and create music.’’
Morreale said it was inevitable the streaming services would be flooded with AI generated music, whether the producers would disclose it or not – and there’s already some on streaming platforms.
‘‘So these models that create new music with AI systems that are trained with millions of songs.
‘‘What happens is that they take this music, compose it into basically numbers and then feed this neural network – a statistical model – and the output of this process is a new song.’’
Exploitation of music was also a big problem, especially for indigenous music such as Mā ori and Pasifika music, he said.
‘‘The original music has some sort of spiritual or ritualistic function other than just entertainment.
Morreale said a lot of people listened to music in the background.
‘‘And so they wouldn’t be able really differentiate, and probably they wouldn’t even care if that is generated by an AI or not.’’
Waikato University student Matthew Burns, who is studying honours in flute performance and music composition, said algorithmic techniques and music composition had been used for over 100 years.
‘‘The core of that is that I am doing something . . . It’s not creating something for me.
‘‘It is an instrument almost through which I am creating.’’
The more commercial AI trained on large pools of data could create a complete song with minimal input from a musician, Burns said.
However, it struggled with making new things, making essential new decisions in new situations.
It could threaten smaller musical jobs – for example composing a short piece for an ad, Burns said.
New Zealand composer and head of Otago School of Performing Arts Anthony Ritchie said the demand could arise from people who wanted to make musicrelated products cheaply and quickly, without having to hire a composer.
‘‘At an economic level, yes, there will be room for AI to be used.’’
As a classical composer,
Ritchie said there was little concern about AI and live performances still had a big place.
‘‘It’s definitely a challenge for those in the music business and we need to remind people that the authentic product is created by a human mind that should be valued, and we shouldn’t loosen up our copyright laws just because of this new technology.’’
But his students should ‘‘certainly’’ be educated on AI.
‘‘We can’t really speak for the distant future, but I think in the near future . . . hopefully AI can be used as a good tool, but not more than that.’’