👋 Hi, I'm Nicholas Roberts. I create and perform music and write this daily blog about creativity, culture, and my life.

I live in Los Angeles with my wife and golden retriever.

Email me: hello@nicholasroberts.io

Subscribe

Get weekly updates of the most popular posts.

The chart below shows the number of Google searches over the last 5 years for "AI music."

If you're a creator of any kind, you've undoubtedly heard that AI is coming for your job. It will soon replace what you do and do it faster and cheaper than you could imagine.

Musicians have felt the impacts more slowly than other creative fields, but that’s changing.

In the last few months, I’ve tried and tested a few AI music-creation tools.

Voice cloning apps let you create pitch-perfect vocals in a matter of seconds, cloning and emulating the world's biggest pop singers.

Splice, a library of royalty-free musical elements, has an AI tool that automatically pairs random sounds together to create new songs and beats.

These tools are wildly impressive. They both harness the power of generative AI and generate new songs or musical ideas based on the patterns of popular songs and what's most likely to come next.

Photographers are competing with deepfake photographs. Filmmakers are up against AI stock footage and storyboarding.

At the time of writing, I'm getting ready for a gig and I'm learning a few new cover songs to add to my setlist. Specifically, I’m trying to learn “How Sweet it Is” by James Taylor.

I've watched about five tutorials on YouTube to show me how to play it. I even watched some live performances of James from the 70s.

But every version I create when I start singing just isn't quite the same as the original. It's good, if I do say so myself, but it's definitively different. It's derivative.

But I'm not playing this cover in my set to be remembered. I don’t want you to comment on how well I played it. Or how similar to Taylor's original recording I can make it sound.

In fact, I too will probably forget the words and chords until the next time I decide to add it to a setlist. And I’ll probably mess up some of the chords.

I'm playing it to connect. Some good friends are coming to my gig and I know they'll appreciate hearing that song, even if it's a rendition.

Talent isn't as impressive as the story you're trying to tell.

Generative AI is talented. But it's not telling a story. It's not building a community.

Great songs come from an emotional and personal place. They're catchy to sing, yes, but they resonate on a deeper human level.

When you're moved by hearing a song you love at a concert, it has nothing to do with how well it's performed. How pitch-perfect it is or even how clever the lyrics are.

Even the strobe lights flashing to the beat of the song can’t replace the fundamentals.

You're moved because of the deeper connection you feel with the performer. The connect you feel with their humanity.

Generative AI is going to wipe out tens of thousands of creative jobs in the next few years. In fact, it already is.

But it can never take away the artistic ones.