Co-creation and ownership for AI radio

Artificial.fm is a 4-year-old academic proof-of-concept with zero stars, zero forks, and zero development velocity—indicating it was published as a paper artifact and never evolved into a functional project with users. The concept (blending AI music generation + user feedback loops + personalization) is a reasonable novel combination of existing techniques, but execution is purely experimental. The platform idea is lightweight and has no network effects, data gravity, or switching costs. Frontier labs (OpenAI/Anthropic/Google) have already moved far beyond this—they have native music generation (Suno, Udio, Google MusicLM) with vastly superior quality, and adding user-feedback steering is trivial product work, not a moat. The paper's contribution is methodological (studying co-creation workflows) rather than technical. The reference implementation appears abandoned and would require complete rebuilding with modern models. Defensibility is minimal: this is a thin application layer on top of commodity music generation + standard recommender patterns. Any serious music platform would integrate this as a feature, not adopt this codebase.