The term “AI” has become something of a boogeyman across industries. Every week, we’re inundated with articles proclaiming the inevitable death of <insert industry here>, but the debate becomes especially precarious when it comes to the implications of creative AI in music. On that side of the aisle, I can understand the anxieties expressed by artists and consumers alike: what value does human creativity and ingenuity hold in the face of algorithms that can generate entire songs in seconds? How could a machine possibly understand the intricate nuances of the human experience?
For several in the pro-AI camp, those questions are moot. Creativity is a lesser concern when money and the opportunity to further develop such technologies in the arts sector are in the picture. Moreover, those same people will (correctly) point out that AI is already widely used in the industry, whether consumers are aware of it or not, and that we simply must get used to the reality that AI will eventually “outgrow” us.
As a fence-sitter who sees both sides of the great AI-in-music debate, I believe that AI should provide the tools to guide musical direction–but never take over it. It’s a delicate middle for such a disruptive technology, but I believe it can be done.
I’ll start by addressing the notion that AI will “destroy” the careers of artists and producers in the music industry: I highly, highly, doubt it. As I’ve mentioned, AI is already being used in music, just not in a robots-taking-over-the-world kind of way (yet). Over 10 different music AI models have already been released in 2023 thanks to companies like Google and ByteDance, allowing anyone to generate songs based on simple text inputs or music samples. This gives a great jumping point for people–both famous and not–looking for inspiration or guidance.
What’s more, many music developers who use AI are collaborating with artists to improve their workflows. With the use of emerging technologies like Spawning, music teams are coming up with new ways to license, monetize, and protect brands with AI models that prioritize artist consent and collaboration with developers. The goal of such projects is to aid in artist development, not create machine-generated Top 10 hits.
However, there’s a limit for everything. A part of me fears that the growing pressure on record labels to constantly produce viral hits will lead to an overdependence on AI to not only create music entirely from preexisting samples (and there’s something to be said about the industry already being sample-saturated), but to mimic real human voices, too. That could lead to a decline in genuinely good artists, or artists who simply aren’t getting paid their work’s worth. While I don’t think it’ll “kill” the industry by any means, producers must be careful not to let mediocrity take over for the sake of shareholder interests.
Another con to the growing use of AI in music is the impending flood of music that is distinctively AI. And I mean that pejoratively. It’s no secret that low-effort, “easy” AI projects are generally uninspired, lazy, and just plain cliché. Think repetitive lo-fi beats and corporate pop, but not even as catchy. “Good enough” music has been growing in recent years, and AI is probably only going to accelerate this.
Regardless of opinion on how involved AI should be in the music industry, it’s clear that AI currently serves as an important and convenient tool in the musical creation process. However, producers and artists must retain a healthy degree of caution with this ever-growing technology to avoid a cultural decline in mainstream music.
