Imagine a world where music—every beat, every note, every harmony—can be generated by artificial intelligence. It’s a world that feels closer than ever, as AI strides confidently into realms we thought were unique to human creativity. Yet, while generative AI has been embraced in art and speech, when it comes to music, something holds us back.

So, why is music, something AI is ostensibly perfect for—lyrics are language, melody and structure are patterns and math—not as easily accepted?

To understand this hesitation, we need to look at the concept of likeness. Think of EA Sports and their use of college athletes’ likenesses. The ethical dilemma is clear: athletes, often unpaid, feel their unique identities are being monetized without recognition or compensation. Now, bring that back to music. If an AI creates a melody that sounds remarkably like the Beatles, where do we draw the line between homage and exploitation? Is it fine for music to “sound” like the Beatles, given they practically invented a genre? Or, is there a line crossed when AI-generated music isn’t just inspired by John Lennon but presented as if Lennon himself is performing?

The fear, I think, stems from history. Musicians have been exploited for centuries—underpaid, manipulated, owned by labels. And this mistrust? It lives on, as we confront a technology that might cut them out altogether. What if, for example, a label, owning the master recordings and rights to an artist’s work, decides to use AI to generate music in that exact style? Legally, they’d be within their rights. But ethically? That’s murky territory.

Yet, there’s undeniable potential here too. Imagine using AI to take lo-fi demo recordings and transform them into high-quality releases. It’s almost like colorizing black-and-white films—a delicate process that raises questions of preservation versus reinvention. Is it honoring the art or overwriting its original essence?

And here’s my last point: generative AI in music is probably a generational thing. What we see as “fake” or “inauthentic” now may be perfectly accepted twenty years from now. Look back to Vanilla Ice sampling Queen or the backlash against Auto-Tune in the early 2000s. Both practices were initially scorned, yet today, they’re widely accepted. Generative AI in music may follow a similar path.

The challenge and the opportunity lie in our response. Will we find ways to honor the artist’s likeness without appropriating it? Can we innovate without erasing authenticity? Only time—and, perhaps, a new generation—will tell.