Sonic Detritus

[Dr. Nelson Harrison]

Deep Fake on the Make…
Reflections on streaming services and AI-generated music
by Liam Noble
21 March 2026

Screenshot of material from Apple Music, 21 Mar 2026 (*)
We should be celebrating. At last, the menial slog and endless tedium of making art has finally been handed over to machines, ending thousands of years of pointless and repetitive graft by humans ill-suited to the task. When I asked ChatGPT about it, they said it was a “practical, economic and creative” move. Like the population in Vince Gilligan’s “Pluribus” , Chat GPT speaks and thinks with one, omnipotent mind that is the product of global online consensus. It’s God, manifested in zeros and ones, an all-seeing, all-knowing and perhaps also, in the fullness of time, a vengeful God? (I haven’t read the Old Testament, but I gather He is not the angel we think he is).

In the meantime, in a bid to hoover up every search imaginable, AI-generated music is now posting on streaming sites using the names of real people. One collaboration, for instance, between pianist Geri Allen, renowned producer Tommy LiPuma (both sadly already passed) and JazzySoul (born in 2025), seemed promising but turned out to be scraps of disembodied cadences glued to a groove that balanced soporific boredom in the execution with simmering rage in the listener. These track credits are now acting as a click-baited portal to the dead-end sludge of coffee suckling backdrops and other gaseous pap that serves simply to fill the once cherished empty air no longer seen as adequate for our restless, dopamine-addled brains. The coffee costs more than the music ever will.


Screenshot from Spotify, 21 March 2026 (*)
Of course, it’s AI’s job to imitate humans, but faster and more efficiently. It can scroll through a selection of short cliches, but it’s like watching a bear riding a bike. You ask a bear to ride a bike and it does it but, putting animal cruelty to one side for a moment, a bear can’t win a race against a human…and AI can’t write melodies, make cadences in the right place or finish a phrase with any kind of conviction. But a bear can…a bear can…( let me just look this up)…it can hibernate, find food with its nose and eat almost anything. By the same token, AI is good at CVs and job applications, which are two things nobody ever enjoyed writing or reading. We are not, as far as I’m aware, lamenting the loss of excitement at delegating bullet points to a bot. Horses, as my nan always said, for courses.

Music needs people. We want to experience it. As listeners, we want to see the humans on stage change emotionally, physically and psychologically. When we listened to recordings, live performances used to be fresh in our minds: we could project those memories on to what we heard. Recordings were a hyperreal facsimile of the moment. Gradually, with electronic music, the need for the performer withered away, and with ambient music we arrived at music as atmospheric decoration, an idea not a million miles away from the kind of stuff Mozart had to trot out for his patrons (and don’t tell me it’s all great…much as I love everyone’s favourite genius, I’ve dragged a reluctant clarinet through enough of his wind serenades and divertimenti to know better). I am a big fan of ambient music, it wears its function on its sleeve, washing over us in a way that is cleverly crafted to not engage or distract. But we are getting lazy, shoving everything into the category of “background”. And people that are making the money are getting greedy, wanting more content more quickly and happy to take the hit on quality. When we hear a gentle guitar and brushes on drums we think “bossa nova” and switch off while the music shuffles from chord to chord like a robot vacuum cleaner on an ice rink.

We are losing sight of the process of making, and listening to, music. AI is concerned with quick and plentiful turnover, with results. But this final stage of music making is only the glittering trophy held aloft for the cameras…what of the pre-match training, the camaraderie of the dressing room, the support of the fans? We are left, like King Midas, cursing our gift, wanting to turn everything into gold and reply to emails with a single click. Now all the food is metal, and the music is a series of wafting, odourless farts in an empty room.

AI learns from us, but are we emulating it too? The countless actions and events per second that computers can perform have rubbed off on us, with our tabs constantly open in multiple apps on multiple devices. And it doesn’t, apparently, suit us very well. Even the old myth of “multi-tasking” turns out to be a fallacy: I walked into a tree the other day because I was searching for something on my phone and failed to see the 30 foot arboreal edifice behind a little black box. If we are trying to work like computers, we are failing, triumphantly, importantly, completely…we are failing. We should stick to what we are good at. If an AI singer and a real one are indistinguishable, two things need to change.

The first is that human music needs to stop modelling itself on zeros and ones and needs to get weird, even just a little bit. Genuine art might have only the barest breath of change in an otherwise predictable landscape…an irrational modulation, a crack in the voice, a kink in the line, might be enough. But there’s always something. Be human. Outwit the competition.

Secondly, people need to reconnect with each other. Go and see music. Feel the air move. Streaming is not dead, not yet, but there’s hope that gig attendance is on the rise again and that people are tired of passable noise leaking from their phones. AI aims at mere competence, but if anything, the results have shown the limitations of that ever-teachable, assessable quality in art. It’s dead, like a motor with no ignition. It needs the spark of inspiration, of desperation. It needs a consciousness that fears death and craves life, whatever that is.

I admit I will miss streaming when it collapses. I’ll miss the instant access to something that crossed my mind and I need to hear immediately, but then it hasn’t made me any smarter or any happier to have that luxury. Somehow the knowledge will not stick because it’s immediately replaced by the need for the next thing.

And having to sift through fictitious, fraudulent and disrespectful catalogues of sonic detritus to get to real music is work that I don’t want to have to do. It’s donkey work, tedious and beneath me, frankly. I would have had some servants to do this in the old days.

Maybe I can ask AI to do it.

(*) Screenshot images used under the fair dealing provisions of the 1988 UK Copyright Act , Sections 29 and 30