Generative artificial intelligence reduces social welfare through model collapse
Fabian Baumann, Erol Akçay, Joshua B. PlotkinUniversity of Pennsylvania · 2026
Read Generative artificial intelligence reduces social welfare through model collapse by Fabian Baumann, Erol Akçay, Joshua B. PlotkinUniversity of Pennsylvania (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-21853.txt.
Description
Baumann, Akçay, and Plotkin model how generative AI can reduce social welfare through model collapse. When models train on data from earlier versions of themselves, outputs lose diversity and accuracy. Delegating tasks to generative AI may pay off for the individual in the short term while widespread adoption degrades future performance, creating a social dilemma. Welfare outcomes follow two dimensions: the incentive to do the task without AI, and the severity of collapse. Habit formation can also spill adoption from low-stakes tasks into high-value ones, amplifying losses.
Catalog record
| CREATOR | Fabian Baumann, Erol Akçay, Joshua B. PlotkinUniversity of Pennsylvania |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58376 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.21853