Op-Ed: Record labels suing AI for copyright infringement – A big deal but all about biz, not music.

Top US labels are suing AI song generators Suno and Udio for copyright infringement based on what you’d call core AI functions. AI synthesizes songs using materials “rephrased” into new songs. The songs aren’t really new in any creative sense, just recycled in many ways.

This admittedly big move comes after much whining by labels and artists about the threat of AI to artists and their livelihoods. They’re terrified of AI, with some reason.

So far, the legal story is that AI-created content can’t be copyrighted. That’s actually still partly up in the air without much case law. That’s a pretty big and useful step, but this lawsuit is very much to the point. This is more about the way in which AI generates music, and it is very relevant.

A bit of background is required here.

Current generation AI is incredibly primitive. It can only “create” based on the work of real artists. It uses prompts to put together a mix of tracks as a recording. This is more like musical LEGO than anything else, and it’s how a lot of hip-hop tracks are mixed, but you can see where AI crosses the line.

It’s not often the music industry (which I despise with a level of sincerity beyond description) does anything right, but this lawsuit really is important. This will define the status of AI-generated content and what’s doable and what’s not.

There’s also the little point that AI doesn’t have to copy anything anyway.

For example:

You can “assemble” your own self-made sounds and wave forms, at whatever MHz, pitch, and tone; literally, you make all the notes yourself. (AI doesn’t really do that. It usually imitates whatever it’s learned.)

You can make your own patches and tracks from your own materials. You don’t even need to sample anything. (That can be good because re-recorded sounds can be pretty awful.)

You still have to be in the ballpark for actual musical sounds, obviously, and not use melodies or other content for copyright purposes, but AI can do it all. That’s without infringing anyone’s copyright and taking food out of their mouths.

Then just do random phrases as you like with what you’ve made, and you’ll eventually come up with something that is actually new and creative. It’s called songwriting, and may even come back in style one day.

This is so incredibly basic. You can make music on just about any level of tech you have available, as long as it’s your own music.

The other word for AI-generated music is “lazy”. Any fool can do it, and some fools apparently do. It’s as bad as formula songs, which are as predictable as stale white bread. No effort is made, and that’s exactly what it sounds like.

It also can’t possibly be anywhere near as much fun as actually playing. It has all the spontaneity of a house brick, and from what I’ve heard, it sounds like a house brick.

Every time I hear AI-generated music, I know what it is, where it came from, how old it is, (that’s horrifying), and how little it does musically. It’s like The Smithsonian’s Greatest Fossilized Hits, but the Smithsonian at least has something to say for itself and a reason for its existence.

The pity of it is this lawsuit is all about business. It’d be nice to define the musical issues, which can be defined with extremely accurate metrics, but not much in the music industry is actually about music. That has to change.

Don’t worry about AI music so much. Do your own stuff, and you don’t need to even think about it. Just watch what happens with this lawsuit, because it is so important.