Greener Than Humans? Environmental Attitudes in Large Language Models
Stefanie Kunkel, Tilman Hartwig, Marcus Voss, Emma K. Schütt, Angelika Gellrich · 2026
Read Greener Than Humans? Environmental Attitudes in Large Language Models by Stefanie Kunkel, Tilman Hartwig, Marcus Voss, Emma K. Schütt, Angelika Gellrich (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-2606-02741.txt.
Description
Kunkel and co-authors test the environmental attitudes that language models show. Such models now help with sustainability choices, reporting, and public messaging. Systematic evidence on their green leanings has been thin. The test draws on established environmental awareness surveys. Behavioral measures join the question set. Thirty-one widely used models are tested, proprietary and open-weight alike. Answers are compared among models and against German human survey data.
Many models lean greener than the average respondent. They show higher environmental affect and cognition. They recommend behaviors with large carbon reduction potential. No systematic link ties green answers to model origin, size, or release context. Prompting conditions matter, though. Persona prompts shift answers in clear ways. Models also mirror user-stated ideology, in a sycophantic shift. That steerability raises concerns for real use. The framework can be reused for value checks. Oversight is urged as models enter public decisions.
Catalog record
| CREATOR | Stefanie Kunkel, Tilman Hartwig, Marcus Voss, Emma K. Schütt, Angelika Gellrich |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.02741