This week delivered two distinct cultural provocations about AI authorship, separated by genre but identical in their underlying anxiety. Spotify and UMG launched AI remix tools letting fans algorithmically reinterpret their favorite artists' tracks. Simultaneously, Granta and the Commonwealth Short Story Prize found themselves unprepared for AI-generated fiction submissions, with no coherent policy to handle them. In both cases, the institutions are scrambling. In both cases, the fans and readers are the ones actually asking the hard questions.
The Superfan Problem: Devotion vs. Desecration
There is something psychologically specific about AI remixes of beloved music. The Verge's framing is exactly right: why would you disrespect your favorite artist with an AI remix? Superfandom is fundamentally a parasocial intimacy. The idea that Beyoncé's voice, her specific grain and intention and history, can be fed into a model and reprocessed by any subscriber is not democratization. It is taxidermy. A 2024 paper in Psychology of Music by Hesmondhalgh et al. found that perceived authenticity in music is deeply tied to listeners' sense of relational reciprocity with artists, meaning fans feel betrayed when the uniqueness of an artist's expression is commodified. Spotify's tool doesn't create new music. It creates the feeling of proximity while eliminating everything that made proximity meaningful.
Literary Institutions and the Submission Flood
The Commonwealth Prize situation is structurally similar but institutionally messier. Granta has been publishing regional prize winners since 2012. The assumption baked into that model was that submitting a short story required a human with a story to tell. That assumption is gone. What's notable is that literary institutions have moved slower than music labels on this, perhaps because the economics of literature never developed the same IP-defense infrastructure. The result is a policy vacuum that rewards bad actors and punishes editors. The question facing both industries is not whether AI can produce technically competent output. Clearly it can. The question is whether technical competence was ever the point.