Music Isn’t a Problem That Needs to Be Solved

As someone who builds and teaches AI/ML systems and has been a jazz musician, composer, improviser, and educator for decades, I’ve noticed that Gen AI is becoming a common buzzword for musicians and on social media. Suno CEO Mikey Shulman said in an interview that “it's not really enjoyable to make music now.” Spotify rolled out an AI Playlist feature that lets Premium users generate entire playlists from a single text prompt. It’s obvious that these companies see music as something that can be produced on demand and cut out musicians. These trends are so disconcerting that I feel a strong responsibility to write this article, clear up common misunderstanding, and mitigate fear with fact.

These developments reflect a corporate belief that music is a product that can and should be produced on demand with minimal human skill, even though music isn’t actually broken. This mindset primarily harms working musicians while offering little value for meaningful creative expression.

Why would AI companies race to “fix” music when it’s clearly not broken? More importantly, the people who lose are the musicians, the composers, the producers, and everyone else who actually makes music. That lack of corporate empathy is morally bankrupt.

Generative AI companies like Suno don’t just offer tools. They’re turning music into something akin to fast food. Suno’s own site says it can create “complete, original songs” from “a single text prompt in under a minute,” and that “no experience” is needed. The same page lists “Impatience is a virtue” as a company value.

For art, this statement is the poison pill.

The difficulty of making music is not a bug. That’s the point. You don’t become a composer by skipping the struggle of composition. You do not develop as a producer by avoiding difficult creative decisions. Instant outputs that require no effort may feel satisfying in the moment, but they bypass the very process through which people grow as artists.

The real danger is not artificial intelligence itself. The danger lies in using it as a replacement for the effort that shapes artistic identity. When creation becomes a prompt, it’s easy to outsource not just the work, but the thinking and judgment that come with it. A 2025 study by Michael Gerlich found a strong negative correlation between frequent AI tool use and critical thinking skills.

Human brains build unique neural connections through personal experience like practicing and composing music. The strength of these connections, what neuroscientists call synaptic weight and machine learning calls weights, grows the more we practice, create, and struggle. Creativity introduces variation that helps prevent ideas from becoming uniform and average.

Generative AI works differently. It compresses large amounts of existing music, much of it copyrighted and possibly even your own, into statistical patterns. When you prompt it, the model generates new output by sampling from those averages rather than from actual musical experience or individual musicianship.

This process accelerates what biologist Richard Dawkins called memetic evolution, the spread of ideas, styles, and sounds through imitation. When new music is increasingly generated from the same statistical averages instead of diverse human experience, musical ideas start to lose their variety and begin sounding the same, and not in a good way.

The more people rely on these tools, the more feedback loops start to form. AI-generated music creeps into the training data and influences future outputs. This process, known as model collapse, leads to a gradual loss of diversity and originality. A 2025 study of commercial AI music generators, including Suno, found that when prompted for regional or niche genres, the models tend to drift toward mainstream styles, flattening the diversity of the music they produce. Importantly, this doesn’t require everyone to adopt these tools. It begins once a meaningful amount of new music is generated this way.

Human creativity is one of the strongest protections against this type of compression. The solution is not to reject AI entirely. These tools are here to stay. Instead, we should avoid using generative AI for the core act of musical creation. Humans should direct the vision and make the meaningful decisions, while AI serves as a supporting tool for technical tasks, iteration, and exploration. Used this way, the result can be stronger than what either a human or AI could achieve alone.

Against all of that, here’s how I actually use AI.

With Cosmoharmonics, I make music that I could only create with the help of AI. It’s the kind of music I dreamed of making when I was younger. I use AI to help bring ideas to life, but I still make the music and the creative decisions myself. That’s also why I started Barrios AI and wrote the book Artificial Intelligence for Musicians. I didn’t write it to give people shortcuts. I wrote it to give musicians tools that help them go further than they could on their own.

When teaching AI and Machine Learning for CBT Nuggets, I keep coming back to one simple quote from kache (@yacineMTB): “you can outsource your thinking, but you cannot outsource your understanding.” That same principle applies to creativity. If you only use prompts to make music, you’re handing over your creativity. Over time, that can flatten what you make. And at that point, the least interesting part of AI-generated music is the person doing the prompting. That’s you.