Music Industry’s AI Dilemma

In India, other than cricket and celebrity weddings, the only thing that can fill up a stadium is a music concert.

Such is the power of music. A song can throw you into depression. Another song can pull you out of one.

You can watch your favourite movie, maybe, 10 times. 50 times. 100 times. But after that, it gets boring. Your favourite song, though? You can listen to it a million times. And the more you listen to it, the more you love it.

And yet—music is, by far, the smallest sub-segment of the media and entertainment (M&E) industry in monetary terms (less than 2% of the entire M&E industry in India). In a country that streams billions of songs a year, the thing that moves us the most earns among the least.

Recent artificial intelligence (AI) developments complicate the story further, by enabling everyone to be a creator and yet further consolidating the monetisation power into the hands of a select few.

Is it then time to realise that the value of music is beyond just money?

Machine In the Song

Recently, Australian authorities decided that a song must be “substantially human-made” to feature on music charts. Purely AI-generated recordings are out.

The trigger was that an AI-assisted cover of “Like a Prayer” racked up nearly 50 million streams—and heavy radio play—before most people realised a machine had a hand in it.

A neat rule. A noble intent. But almost impossible to enforce fairly.

Because tech and AI are already everywhere in music. In the arrangement. In the mixing. In the mastering. In the chord progressions. In the singing (AutoTune, anyone?). Even in ideation—Grammy-winning songwriters use it to spark ideas.

Ironically, those songs will get away without the AI tag.

But at the other extreme, an original song in which the creator wrote every lyric and melody themselves, and used AI only to produce it gets flagged, because of the embedded signatures.

So where do you draw the line between a tool that helps a human create, and a machine that creates instead of the human?

Nobody has a clean answer yet. Maybe we are asking the wrong question. Maybe it was never about how human the output was. Maybe it is about consent at the input stage. A teenager in a bedroom learning Jimi Hendrix riffs is inspiration. A model trained on Hendrix is something else entirely. Same source. Very different act.

Cher’s “Believe” was one of the most machine-altered voices in pop history, and still, unmistakably, human. Because a person brought the intent, the emotion, the lived experience. The machine only rearranged what it was fed.

Imitate, yes. Create? NO

I write songs. And I can tell you this: a model can imitate a thousand heartbreaks. It cannot feel one. It has never waited for a call that never came. It has never turned a private ache into a line that a stranger somewhere calls their own.

That is the part that goes beyond money. And that is the part worth protecting.

But let us not play Luddite either.

When samplers first arrived, the industry panicked about stolen intellectual property too. That same technology went on to birth hip-hop and electronic music. AI can be that kind of force, if we set the rules while we still can.

So, opt-in training, not opt-out

Rights for likeness and voice. Honest disclosure of what a machine made and what a human did. Set them well, and creation opens up. Set them badly, and power only consolidates further.

And the market is already moving. After a wave of lawsuits, Suno and Udio, two of the world’s biggest AI music platforms, have signed licensing deals with major labels and are releasing newer models trained on licensed catalogues.

Progress, of a kind. But read the fine print. Suno’s own terms have subtly shifted: a paying user who “creates” a song on Suno may now use it commercially yet generally does not own it, because a machine generated it.

Sit with that. The one thing the human is denied is authorship. That is the whole debate, in a single clause.

So, back to where we began: Music has always been undervalued in rupees and dollars—perhaps because its real worth was never measured in either.

AI, by threatening to commoditise the output, is forcing us to finally name what we were undervaluing all along. The human hand. The consent. The lived experience. The creators’ voice.

Protecting music in the age of AI is not, at its heart, a royalty problem. It is a values problem.

A stadium does not fill up for a machine. It fills up for a human’s experience—something a machine can never be. Such is the power of music. Period.

(The author leads the M&E industry practice for a leading management consulting firm and is an award-winning lyricist. Views expressed are personal.)