Music AI has been around a long time, it's just being sold to you differently now

Hey there!

I keep seeing conversations surrounding AI and music pop up over, and over, and over again. It's understandable. AI is not going away any time soon and its relationship with music will continue to be fluid for the foreseeable future.

While there has been a lot of news recently, like major labels settling with Gen AI startups and somebody buying an AI song's way to the top of the country charts, what I want to chat with you today is about the difference between generative AI and assistive AI and how you've likely been using one a lot longer than you might realize. And, also why that's entirely okay.

AI is just machine learning

I have two degrees in creative technologies and back when I was in school a decade ago, a lot of the exciting conversations were about the creative implications of machine learning. While it isn't my area of expertise, I've always been fascinated by it. Machine learning is essentially a complex algorithm that is able to make a prediction or decision without being explicitly programmed. This is achieved by feeding it data and getting it to recognize patterns, anomalies, and so on.

What we call AI today is just machine learning with a marketing facelift to appear more sexy and dangerous.

If you understand machine learning in its most summarized form, it's easy to grasp why so many musicians (and other artists) are enraged by generative AI in particular. AI can't exist without first being trained on the works of these artists. Without consent and compensation, generative AI is unethical.

But what about assistive AI?

Aside from the legal and ethical problems that come with generative AI, it doesn't produce work good enough to meet professional standards in any real way. This is the exact opposite of assistive AI, which is pretty uncontroversial and actually useful for professional producers.

Part of the reason why assistive AI isn't a hotbed of controversy is because it's been a part of music tools for years now. Izotope's Ozone Mastering Assist feature is assistive AI, for example. Nobody predicted the collapse of the music industry when that feature was introduced in 2017. It also didn't require anybody's work be stolen.

This is the future of music AI worth caring about: assistive tools that help hobbyists and professionals alike speed up their workflow without hiding everything in a blackbox. These tools are often associated with mixing, like Ozone, but they can also creative as well like XLN Audio's LIFE. These tools often run locally on your system (so no dolphins are being boiled alive to cool a data center somewhere) and don't steal your work without consent. They don't create legal risks for you or your clients.

The distinction is important

It's important that we're able to know and articulate these differences between generative AI and assistive AI. Startups are going to continue to attempt to force feed us novelty generative AI tools that take away your agency, creativity, and rights. We should all demand better tools and that starts with knowing what we're asking for.

Til next time,

ZW