If I were to ask any LLM to write an article for me to say goodbye to the year 2025, it might have got one thing right: the article would have centred around AI.
This year, my focus was not on my intelligence but on that of the widely unknown machine that is silently creeping into our lives. In some areas, it is welcome. But in some, I want to be the scary, old-aged gatekeeper screaming at people trying to smuggle AI-powered chatbots into my area of interest, which is entertainment.
A few months back, I came across several reports on the use of AI in entertainment. One that caught my attention (and made me a little obsessive about the topic) was a report on AI bands and their impact on the music business. The song ‘Dust on the Wind’ (not to be confused with Kansas’s Dust in the Wind) from a seemingly AI-generated psych-rock band ‘The Velvet Sundown’ has so far registered hundreds of thousands of plays on streaming giant Spotify. Now, some reports suggest we have reached a point where it is difficult to tell who made the music (humans or AI).
In one of the reports published in AFP, I came across a quote from a music composer who said that “AI highlights the chasm between music people listen to “passively” while doing other things and “active” listening in which fans care about what artists convey.” My objection to this reasoning is that a piece of art only becomes valuable when its admirers are aware of how it is created. There is no ‘passive’ listening, in my opinion. Timeless classics carry soulful meaning for listeners who pass it on to the next generation, making art and music immortal.
If we were to understand how models might have been trained to create music, we can assume they were fed music libraries to learn patterns and ‘predict’ the next best possible outcome. That’s how they come up with a tune – whether it’s forgetful or not is another case. Doing so would mean that we are restricting the arts (which include music, writing, singing, dancing or any form of creative expression) to labelled boxes, that we are acknowledging that the arts is supposed to follow a pattern.
The very nature of art is rebellious, and that’s what makes it both exciting and controversial. Art cannot (and should not) follow or appease the status quo; its questioning nature is what separates it from mundanity. Many moons back, I read a paper on subliminal messaging in songs. It said that in the days when rock bands were accused of promoting Satanic messaging (Led Zeppelin’s Stairway to Heaven, for example, was blamed for having objectionable verses when played backwards), some band members started purposefully adding notorious messages just to tease conspiracy theorists.
The new wave of ‘AI music’ got me thinking: will we ever have musicians and artists courageous enough to give back to troublemakers? Does a model have the ability to handle controversies and add its own touch to it?
My favourite part of music is the many behind-the-scenes (BTS) videos that come with them, where artists share how the melody came to life. The story behind a song makes it memorable and becomes part of what people now say ‘active listening’. With AI, will there be such stories? I fear not.
My issue with such models is that there is no mechanism for explaining why a particular output was generated. How a model’s neural network behaves and why so. With humans, there is logic in no reasoning as well. ‘Flop music’ has someone holding themselves accountable. With AI models, there’s silence (at least so far).
In my years of listening to music (in the world of Gen Z, I often see myself as an extremely irritated bespectacled woman helplessly holding on to the glorious days of the past), I have learnt one thing: every individual has their own way of connecting with a song. For me, Queen’s ‘Show Must Go On’ is therapeutic. I used to listen to it on repeat when I was at my lowest low. Besides that, the fact that this was one of Freddie Mercury’s last songs makes it even more meaningful. It shows the passion of an artist who chose immortality through his work over the silent last days, battling an illness known only to a few. When I listen to it, I imagine as if Freddie is still alive, watching us with pride as we enjoy his music.
One of the stark celebrity deaths I have encountered is that of actor Robin Williams. The news that he decided to end his life was even more traumatic. He was the person who, when Christopher Reeves was in the hospital after a tragic accident that left him paralysed, brought a smile to his face. And he was the person who somehow thought there was no purpose left in living (after his diagnosis of dementia). Iron Maiden released a tribute song for him shortly after his passing. ‘Tears of a Clown’ has a haunting line: who motivates the motivator? And every time the song appears on my playlist, it becomes hard to ignore it. When I was listening to ‘Dust on the Wind’, I kept wondering if my near-disgust for the song was because it lacked its soul. Will we always need humans to blow life into songs (through catchy tunes and memorable lyrics)?
The funny thing is that in my Spotify Wrapped or YouTube Review, these songs may not be on the top. So, if an AI model were to learn my listening habits, it would have trouble understanding what songs speak to me, mostly because it wouldn’t have sufficient data to predict my preferences. The songs I used to like came from a time when listeners had fewer options and no immediate apps to stream them. We used to scour websites to download songs or transfer them to our phones through laptops or computers.
The human brain is fascinating. It does not necessarily have to see a particular data set countless times to make a pattern. On a personal level, there is a song that I may not have listened to on my own. It’s a qawwali by Aziz Nazan titled ‘Jhoom Barabar’; this is my father’s favourite and on weekends, I would wake up to the singer’s powerful voice blasting from the giant speakers placed in our TV lounge, barely concealing my dislike for the song that was clearly different from those in Disney movies we were allowed to watch. And yet, the adult me brain has a deep connection with it. In this case, passive listening helped me connect with a song, but that was only possible because of someone else’s active listening. And that’s how art keeps getting a new lease on life.
Art defies a rules-based order and makes people think. It is what makes social and political commentaries possible. Music is dependent on artists (who think and feel differently) and not on GPU-powered AI models. Machines can learn the pattern of an artist’s work and predict on it, but to say that it can create a connection with people is a little unbelievable, at least for me.
In the world of likes and virality, an AI-powered band can make its space on streaming platforms, but there will always be resistance from ‘active’ listeners. Music for passive listening is nothing but glorified noise, and I can only hope that all those who admire music would do something to stop music (human-made, made for active listening, whatever one may call it) from dying.
