From Barking Hellhounds to AI Slop: What Electronic Music Foretells About GenAI (words.narain.io)

OpenAI’s Sora and Meta’s Vibes — apps for instant AI video creation and remixing — mark a “ChatGPT for creativity” moment, but they also crystallize a long-standing fear: the rise of “AI slop,” an avalanche of algorithmically produced, low-attention content. The piece traces this anxiety back to 1950s complaints about alien electronic sounds and follows democratization through MIDI, cheap synths, and DAWs to Moby’s 1999 album Play, whose ubiquity as background music helped normalize content optimized for utility rather than deep engagement. Today that model collides with scale: roughly 120,000 new tracks are uploaded daily and 87% of 200 million tracks don’t reach Spotify’s 1,000-stream payout threshold, reinforcing a power-law “superstar” distribution rather than a thriving middle. Technically, GenAI is the logical extreme of these trends: it makes generation trivial and multiplies content supply, while Sam Altman’s warning about an “RL-optimized slop feed” highlights how reinforcement-learned recommendation systems can amplify mediocrity if optimized only for short-term metrics. The crucial shift is from building better generators to engineering selection systems — algorithmic gatekeepers, provenance and quality signals, incentive design and human-in-the-loop curation — where the embedded values of filters will determine what content surfaces and what culture becomes.