AI-Assisted Stories Score Higher but All Sound Alike, Science Advances Study Finds
lmoroney · x · 2026-10-13
A Science Advances study by Anil Doshi and Oliver Hauser had 293 people write eight-sentence stories; some could request up to five three-sentence starting ideas from GPT-4, and 600 blind readers rated the results.
- Individually better: AI-assisted stories were rated more novel, better written, and more enjoyable. Those allowed five ideas scored 8% higher on novelty than solo writers, with the lowest-creativity scorers gaining the most—roughly catching up to the top group.
- Collectively homogenized: AI-assisted stories were more similar to one another. The authors frame this as a social dilemma: each writer is better off, but the collection as a whole narrows.
- Notably, no bad behavior is required—each writer taking a good idea is sensible, yet a thousand sensible writers still produce a shelf of alike-sounding books. Writers' own satisfaction didn't differ across conditions, so both the gain and the loss are hard to see from the individual's seat.
Related event: GPT-4 Boosts Story Novelty but Makes Writing More Homogeneous(2 posts)→
More from AGI Musings
- Sakana AI proposes MASS: multi-agent self-supervision for recursive self-improvement — hardmaru · 2026-10-13
- Dan Shipper: cracks are appearing in the frontier model duopoly — danshipper · 2026-10-13
- Stop chasing AI frameworks: dev says spend 6 months on the math instead — TivadarDanka · 2026-10-13
- Anthropic researcher recounts conversion: frontier insiders converge on same AGI conclusions — abhiadesai · 2026-10-13
- Esther Dyson's new book 'In Defense of Limits' asks how humans should govern AI — Grady_Booch · 2026-10-13
- Influencer tells left-leaning audience to follow AI Twitter, the internet's last non-doomer corner — mallow610 · 2026-10-13