Definition
Each person is better off using AI while the collective pool loses novelty, so individually rational adoption produces a collectively worse outcome. In Doshi and Hauser's experiment, 293 writers each wrote a short story with no AI ideas, one or five. Novelty rose 5.4 percent with one idea and 8.1 percent with five. The stories also became measurably more similar to one another.
What is at stake
Nobody in the room has a reason to stop. The gain is real and it is theirs. The loss is spread across everyone and appears in nobody's metric, which is the shape of every commons problem, and the reason a policy cannot rest on individual judgment alone.
Where the word came from
- The tragedy of the commonsGarrett Hardin, 1968
- Governing the commonsElinor Ostrom, 1990, the evidence that the tragedy is not inevitable
- Creativity as a social dilemmaDoshi and Hauser, 2024, the frame applied to AI-assisted writing
How episcope came to it
Anil Doshi, UCL, and Oliver Hauser, University of Exeter, in Science Advances, July 2024, a peer-reviewed journal. The experiment was pre-registered, with a separate pool of 600 evaluators rating the stories blind, which is what lets the collective claim be measured rather than asserted.
What to watch for
Two limits. The gain went almost entirely to writers who scored low on the creativity pre-test, so the equalising effect is as much the finding as the narrowing, and there is no evidence the ceiling rose. And Hardin's name travels with a caution: his politics were explicitly racist and eugenicist, and Ostrom's work showed his inevitability claim to be wrong. Ostrom is the better ancestor.
Papers and articles behind it
Doshi, A. R., & Hauser, O. P. (2024). Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances, 10(28), eadn5290.