MBW Views is a series of exclusive op/eds from eminent music industry people… with something to say. The following comes from Eamonn Forde (pictured), a long-time music industry journalist, and the author of The Final Days of EMI: Selling the Pig. UK-based Forde’s new book, Leaving The Building: The Lucrative Afterlife of Music Estates, is out now via Omnibus Press.
Music business coverage in consumer publications and websites in recent years would mainly stretch to asking pop stars about streaming payments, but in the past few weeks a new question is being asked. It is about what AI will “mean” for songwriting and music production, as if any of this is brand new and hasn’t been rumbling in the background for years.
It serves, however, as an important staging post: this is no longer an “inside baseball” issue wherein the music industry self-importantly discusses niche music business matters in esoteric terms with itself. This now goes much wider.
For example, just before he became an Ivor Academy Fellow, Sting was asked about AI in songwriting by the BBC.
“Maybe for electronic dance music, it works,” he said, perhaps revealing his own biases about what he believes is and isn’t ‘music’. “But for songs, you know, expressing emotions, I don’t think I will be moved by it.”
He added, “The building blocks of music belong to us, to human beings.”
Sting, like Bono, is sometimes such a ridiculously pompous character that you almost feel duty bound to disagree with every syllable of what he says. But, as with the stopped clock in Withnail & I, he’s onto something this time. The human can create music in a way a machine cannot because they can capture at least some of the intangibility of the human experience.
The recent establishment of The Human Artistry Campaign, even in its nomenclature, stresses a similar point. “AI can never replace human expression and artistry,” it says. Human good; machine bad.
This human/machine dichotomy makes it all an incredibly emotional subject. Machines can only approximate clichéd notions of human emotion; they offer immanence not transcendence.
Songwriting, as Nick Cave so memorably and forcefully put it recently, is “a blood and guts business” that “requires my humanness” in order to happen and to convincingly articulate the chaos and the beauty of being human.
There are wider commercial debates about the problems of AI, especially with its ability to be quickly and cheaply produced purely to game DSP payments. That’s a whole other matter that has been well covered here and here and here. Equally, the hullabaloo around soundalike tracks, symbolised by the recent “Fake Drake” controversy, is another debate as that is something that trademark laws as well as name and likeness rights should be used to police.
“There are wider commercial debates about the problems of AI, especially with its ability to be quickly and cheaply produced purely to game DSP payments.”
What is at risk of getting drowned out here is a qualitative and aesthetic argument around AI-generated music in general – where it becomes a new type of Muzak, or MusAIc as I termed it earlier this year – and if it will ever replace human-created music. In brief, the argument is this: if you are worried about AI music stealing your audiences, what does that say about your faith in your own music?
There are two false premises running in tandem here: 1) that all AI-generated music is inherently bad and wrong; and 2) that all human-created music is inherently good and valid.
With regard to point 2, there are vast amounts of human-created music that is soporiferous and invalid. Pointing this out to songwriters never goes down well. But it’s still true. A lot of music is mediocre and unnecessary. But it needs to exist in order to make the exceptional and the necessary shine. Great art can only be understood and underscored when run in sharp contrast with art that is desperately lacking. We only understand and appreciate the supreme when we have had to chew on grey gristle.
But, and this gets to the heart of the matter, if you are worried that AI-generated music is somehow, in the raw and open market of capitalism, going to be stealing listening time and audience enjoyment that should, rightfully, have gone to your music, that takes an incredibly dim view of both the audience and the quality of your own work.
In the early days of the sociological study of media effects in the late 1930s, there was a model that directed much of the thinking here for a long time: the magic bullet or hypodermic needle theory. In brief, this posited that the media pumps something out and the audience, passive and stupefied, swallows it unquestioningly. Media messages were powerful, unidirectional and accepted.
All of this suggested that the audience, in modern parlance, had no agency. The average audience member was a patsy, a submissive automaton. It had massively overestimated the power of media to get its message across as intended and massively under-estimated the power of the audience to critically dissect and evaluate what they were being presented with. Humans are more complex and active than they were being given credit for.
This kind of thinking has returned and is the subtext of a great deal of the criticism of AI music. Audiences are apparently largely lacking in intelligence and so will simply be hoodwinked by whatever is presented to them. They are supposedly so feeble-minded that they will never grasp the music they are mistakenly dancing to and forming memories around and having their synapses inflamed by is the result of algorithms and machine learning, not of humans expressing the depth, the complexity and the actuality of being human.
The music business at the top end has always had an audience appreciation problem, insisting it knows what is best for the consumer and then finding multiple ways to profit from its self-proclaimed expertise. Sometimes it is right and art has its hand firmly on the tiller; but sometimes it is so ridiculously wrong as to self-satirise.
Be very worried, then, about the rising panic around the perceived industrialisation of the monetisation of AI-generated music; but equally have faith that the consumer can quite easily tell the difference between music that will change their life in a thousand ways and music that just sounds like a karaoke version of itself.
AI technology is good but it’s not that good. Audiences are a lot smarter, and they are far more discerning, than the music business sometimes gives them credit for. If it is true and they are as passive as much dystopian thinking around AI music presumes, surely a lot more of the heavily marketed, but absolutely terrible, records over the years might not have sunk to the bottom of the ocean quite as hard as they did.
If you are worried that audiences are simply not discerning enough and will unthinkingly gobble up any old shit that is pushed in front of them by AI tools, is that because you’ve presumed a business that has been built on audiences unthinkingly gobbling up any old shit made by you?Music Business Worldwide