A human premium emerges as AI slop floods everything
Reports are surfacing of a growing 'human premium' for art made by people, and commenters pointed out that originals have always sold at a premium over identical copies. Older threads connect: a request to add a flag for AI-generated articles on HN (48886741), Cliff Stoll's 'Rumors of my death are slightly exaggerated' after an AI-written obit (48037336), and the new HN rule against posting generated text. One more small data point: a developer paid real people to follow his README and found a missing first step (php-fpm), which one commenter noted used to be called hiring QA.
The pattern: as generated text gets cheap, proof of human attention becomes the scarce thing. People trust the human who got stuck on step zero more than a clean-looking AI summary.
The hard part is enforcement. Commenters note AI detectors are unreliable, voting can be gamed, and heavy anti-LLM stigma drives away people who use these tools lightly.
So what?
If your product involves content, reviews or community, plan for provenance: show who made it, who tested it, who stands behind it. Paying real people to try your product is a cheap, credible signal and still finds bugs no model catches.
Read these
AI is creating a 'human premium' for art created by people
I paid people to try and follow my README
Ask HN: Add flag for AI-generated articles
Rumors of my death are slightly exaggerated