What we lose to AI music

AI-generated music is here to stay, whether we like it or not. If you are anything like me, you'd have a negative reaction to the concept of music made by a machine. However, let's keep an open-mind and explore what lies ahead.

I will not cover copyright or AI ethics or AI-assisted music production. Nor the outright criminal acts that are necessary to create the technology. The music discussed here is Western contemporary music, which includes lyrics.

Before we dive deeper, I want to make the distinction between making music and generating audio files from a prompt. They might both end up as audio files, but these are two different processes.

Additionally, it is an ethnocentric view (and arguably capitalist) that a recorded song is the only expression of music. There are many other cultures that combine dance, poetry and other social aspects into the word 'music'. We have not yet muddied the water there and that gives me comfort.

machine music is not all bad

The most famous benefit of listening to music is that it helps us regulate emotions. Listeners can select tracks to match or shift their mood. They'll reduce stress and have stronger positive feelings through self-soothing and dopamine releases. When it all gets too much, we can retreat into our music.

Listening to music can enhance focus attention, for when you want to get work done. It is used as a motivator for mundane tasks and improve performance in sports through the arousal effect. Under the right circumstances it can even improve memory and aid in language learning.

Logically, all these effects could also be achieved with generated music. And there is no strong evidence this isn't the case.

Adam Neely mentioned that AI music could be helpful in music therapy, in his excellent video essay. I can imagine such applications, as well. We could create songs that remind a person with Alzheimer's of their children's names. I don't see anything 'wrong' with that, per se.

We shouldn't dismiss machine music entirely.

art vs. utilitarian

One mental model I explored is the expressive versus utilitarian music. With utilitarian, I mean music where listening to it is not the primary aim.

We could generate music in real-time during video games, based on what happens in the game. Or create music based on physiological markers–like in focus apps. Another application could be background beats to fill up the silence in a talking-head video. Or themed playlists to set the mood in a restaurant.

Aside from the economic impact on musicians, I can understand the use of generative music in these settings.

However, the music that awes me. That I listen to attentively. That is what I call expressive music. And this art has aspects that can't be replicated with a machine: expressive recognition.

When we listen to artists we are mirroring the emotions in a track. We recognise and empathise with the artist. A key aspect of music appreciation. But, whether consiously or not, we know that the computer never understood the content expressed, so it robs us of that inherently human experience.

The same goes for structural recognition, where we recognise rule‑breaking of genre tropes and musical elements like odd progressions, tempo shifts, and key changes. All moments where the musician signals a shared understanding of cultural and musical understanding. Machines lack this syntactic awareness. They'll never share a virtuosic wink, one that connects us on a deeper, social level.

bypassing the benefits

You could make the argument that these aspects of music are not important to the average music listener. You might be right, although I hope not.

But that's not my main worry.

Generating music through prompting will completely bypass the cognitive, sociological, and psychological benefits of writing and performing music. What if people won't attempt to master instruments? Quit collaborating on, creating and sharing their music? What if they stop learning about music altogether?

People will lose out on music teaching them to be effective communicators and creative problem solvers. They'll lose out on the self-assurance they'd gain from showcasing their mastery of musical instruments.

We'd create a generation of 'musicians' with poorer social emotional functioning, because they never had to compromise on their ideas. They'd dull their emotional intelligence because they won't have to navigate through social conflict.

Without the struggle, 'musicians' will not learn the life skills they would've otherwise.

society and politics

And that's the internal aspects. What about the societal and political aspects of musicking? Music has an dual role in society. It's both associated with order, religious and political power but is also a subversive force.

Music is used as a conduit against the status quo. Against the order. With the homogenization of music (making all recordings sound similar and the music formulaic) we risk losing its role as a balancing medium.

Think of the songs of protest against the leaders of the world's superpowers. They are quick to dismiss the artists, or even prosecute them, knowing and fearing the power of music. Persons that generate music with AI, will not have the capacity, nor the following, to 'rebel'.

We also shouldn't ignore the tons of politically charged, generated songs designed to skew democracies and fortify ideologies. More the reason that human music needs to persevere. We need to keep the balance.

where do we go from here

If we want to weather the storm, our relationship with music has to change. I implore musicians to explore more interesting, perhaps theoretically complex music. Sounds that have not been done before. And music listeners should keep an open-mind to these sounds.

From an economic perspective, we, the music lovers, should make collaborative and live performances more accessible and discoverable.

We should innovate on human curation in an increasingly saturated market. We have the responsibility to pick out the genuinely interesting, expressive music that explores the boundaries.

Let music be a subversive force.

Paste this URL into Mastodon/Fediverse to reply.

Loading comments...