|
|
|
|
|
by delbronski
41 days ago
|
|
I recently used Suno to generate a whole album for myself. I was after a specific style. I had this one song in my head and could not find anything else like it. So I tried Suno, very skeptical of it, but was completely surprised by what it could do. So I spent a whole day making an album for myself. Completely mesmerised and blown away. Even thinking… I should publish this, this is amazing music. The next day I was listening to it, and after a little while the magic had faded. It felt like just a very generic, average crappy album. It didn’t have the lasting effect and replay value that my favorite albums have. Maybe it’s because I knew it was AI. Or maybe the magic came from the tech and not the actual music. Anyways, I gave up my music production dreams after that. |
|
But for music especially there is a very deliberate process a human uses. Musicians looking to create something original are not pumping it out over the course of a day/week/month. It is something they have thought about for years. A lyric waiting for the perfect melody, a melody waiting for the perfect baseline, etc. OK, maybe there is a moment of inspiration where it all comes together quickly, but that doesn't happen in isolation.
I am sceptical an AI with no goal or understanding beyond it's trained weights and prompt can create something original (and good) in it's entirety.
This is in contrast to technical tasks like coding, where regurgitating best practices to solve an iteration on an already solved problem is a valid approach, for everything except bleeding edge. No originality needed in that case.