Few marketing blurbs differentiate between Machine Learning and Artificial Intelligence, so it's hard to know exactly what they're doing. For example:
Machine Learning could go through all the presets you use in live performance and balance their levels. It would "learn" what level each preset uses, compare them, and edit them for the same perceived value.
Artificial Intelligence could go through all the presets you use in live performance, and generate new presets which meet the same general criteria as what you seem to like.
Companies are finding out that AI isn't generating much $$ yet, and there are also legal issues involved. AI has much potential - I used it to generate two recent book covers. It could also be tremendously helpful in reducing repetitive work in the studio. But, so far, practical applications for music that truly use artificial intelligence don't seem to be very common.
Presonally, I'm a fan of artificial stupidity. Some of my best sounds come from making mistakes and thinking "y'know, that actually sounds kind of cool."