I’m going to preface this post by giving a quick background about myself as it’s important to the conclusions in this post.

I work on AI in my day job. I’ve been in tech my whole career and have found myself recently focused on AI as the technology has gained popularity and infiltrated every company.

I work on music at night as my hobby. I’ve released two albums in the past two years. I’ve played guitar for a long time. I’ve recently dedicated a lot of time and effort into learning how to solo-produce music. I love it.

It’s an odd dichotomy I’ve found myself in. I think when used correctly, AI can be a tremendous asset to all people. But what about a world where anybody can type a question to an AI model and get back a fully fleshed out song? Is that really human generated art?

The Anti-AI Argument

Artists of all types are concerned that AI is going to take some of our work away. We’re seeing artists create music videos with AI and we’re seeing other artists criticize it.

It takes a lot of effort to get to a point where you can create good music. All the hours spent learning how to write music, learning your DAW inside and out, creating a good mix, creating a good master… these things take a lot of time to learn. Seeing a computer generate something that’s of a surprisingly good quality in a few seconds is flat out disheartening. More than the effort, however, is the emotional investment put in by musicians. Humans pour their heart and soul into creating a track that represents them - something that AI completely loses.

Imagining a world where AI-created music is ingrained everywhere - movies, commercials, radio - do we start to lose the human-touch of our day to day life? Human generated art is authentic. It’s the result of a person’s mind and hands coming together to make something unique. The thought of having the majority of the art around us be made by a computer is depressing.

But also, how far will it go? Imagine you walk down to your local coffee shop and you order from a screen, have your coffee created by a robot, and have no human interaction - that’s even more depressing.

De-humanizing society is a slippery slope. I think IF we get to a point where majority of art (paintings, imagery, music) is created by AI, we are headed down the wrong path as a society.

But I don’t believe it will go that far.

The Value of AI

The value of AI can’t be denied. Working in tech, the amount of times I use it to help me code, explain concepts, even create PowerPoint content has saved me so much time.

This same concept of using AI to make your job easier should apply to music creators as well. And there are already plenty of AI-Assistants popping up to do this:

• WavTool - Browser-based DAW with tons of AI features

• Magenta Studio - AI tools for Ableton Live

• LALAL.AI - Using AI to extract stems

• Orb Producer - AI computed melodies

But there’s also tools popping up that flat out generate the whole piece of music from a prompt. Take AIVA.AI as an example. Their tagline is “AI Generated Songs in Seconds”. I view this very differently than the tools above that help you make music.

There has to be a line drawn in using AI to assist your music creation workflow vs. using AI to create your music.

Drawing the Line

My opinion is this - being in control of your music workflow but using AI as an assistant is still human-generated art. We should be leveraging the tools we have to help us and I don’t see art as an exception to that rule. These tools will help more people learn how to make quality music, and given I believe music creation is one of the most fulfilling hobbies there is, I’m all for anything that can enable more people to get into this craft.

I do think there is a fine line on what is using AI to assist your workflow vs. what is using AI to create a whole track. Take generating a melody from an AI tool as an example. Doing this and using it in a final product feels pretty disingenuous to me at this point in time.

But honestly, should it? Artists have been taking ideas from others around them for centuries. As AI grows in popularity, it’s going to become more and more ingrained in our day to day lives. So much so that it may become another source of inspiration for artists to draw upon when they sit down to make music.

But to be clear - I am not a fan of anything where the AI is doing all the work to create the song. I’m sure some of you out there may be thinking something along the lines of “prompting the AI is the human interaction part of the workflow and it should count” but I don’t agree.

Writing a description and getting back a song is so different than the normal musician workflow. Compare this to using an assistant to get a melody and integrating that into a larger track where the human is deciding all the ancillary pieces and textures around that melody. They feel very different to me.

Differentiation

One avenue that will continue to increase in popularity as we want to distinguish between human-generated art and AI-generated art is AI content detectors. There are already all sorts of these types of tools popping up in the writing space.

As an example, professors don’t want their students’ papers to be written by ChatGPT, so there has been a boom in these types of tools to flag this. Originality.ai is one example that claims a 99% AI Content Detection Accuracy.

I can definitely see this same type of tooling gaining popularity for music as well. People aren’t going to want to listen to AI generated music. Being able to distinguish what was created from a human vs. an AI model will play a role in the music industry.

In fact, I recently stumbled across a company called Believe (parent company of music distribution platform, TuneCore) who is doing just this.

Notice his distinction of 100% created by AI. That supports my stance above on the distinction between using AI as an assistant in your music workflow vs. having AI create the whole song.

It’s great Believe is heading down this path and expect the major streaming platforms to implement something similar. Spotify isn’t going to want the overhead of having to store and serve all these AI-generated tracks and listeners aren’t going to want them popping up in their recommendations.

Conclusion

AI & Music should be friends. There should be no shame in using AI for learning, efficiency, marketing, and more. People who use AI to create music from a prompt are not musicians and their work shouldn’t be labeled as such.