The easiest story about AI music is that a machine made the song. It is also usually the least interesting one.
British creator Oliver McCann describes himself as a music designer. He writes lyrics, then uses generative systems to work through version after version until a track matches what he had in mind. His breakout song, Stone, turned that process into a record-label deal—and a useful challenge to the idea that authorship begins and ends with traditional performance.
The tool can supply a voice, arrangement and production. It cannot explain why one version should survive while ninety-nine others disappear. That choice is where taste enters the room.
“The prompt is not the finished idea. The edit is where an artist reveals what they value.”
Listen before you litigate
None of this settles the harder questions about training data, consent or compensation. Those questions deserve reporting that is specific rather than reflexive. But they should not prevent criticism from doing its first job: listening closely to what a piece of music actually does.
Silicon Stereo will name the process when it is known, link to reliable credits and distinguish a synthetic persona from an artist using AI inside a larger practice. Transparency is the starting point. The music still has to carry the rest.