Are 2023 AI music generations already becoming historical artifacts?
In 2023, tools like Suno, Riffusion, and AudioCraft were in a very different stage compared to today. The early v0/v1 generations were often strange, imperfect, and experimental — short clips that felt more like prototypes of a new medium than finished songs.
What interests me is not just the technology itself, but the human perspective around that moment in time.
I was an early adopter who used these tools extensively during that period — not just casually experimenting, but being deeply immersed in the process. Over time, I naturally curated what I considered the most interesting, unique, or memorable examples rather than simply keeping every raw generation.
That makes me wonder: does that kind of firsthand curation have historical value?
When early synthesizers, computer music systems, samplers, and electronic instruments first appeared, many of their early experiments were also viewed as strange or disposable. In hindsight, those recordings became important because they documented the first stages of a new creative technology.
AI music may eventually have a similar history. The earliest generations may not be valuable because they are technically impressive by today's standards, but because they capture what it felt like when this technology first became accessible to everyday creators.
AI companies may have enormous internal datasets of generations, but a raw database does not necessarily contain the human context of:
- which generations stood out
- which represented interesting breakthroughs
- what creative approaches produced memorable results
- what early AI music actually felt like to experience at that moment
I'm curious whether early AI music should be preserved as part of music history — similar to early electronic music, computer music experiments, or early internet art — or whether these artifacts will ultimately be viewed as disposable experiments.