A psychometric framework for evaluating and shaping personality traits in large language models
Gregory Serapio-García, Mustafa Safdari, Clément Crepy, Luning Sun, Stephen Fitz, Peter Romero, Marwa Abdulhai, Aleksandra Faust, Maja Matarić · 2025
Read A psychometric framework for evaluating and shaping personality traits in large language models by Gregory Serapio-García, Mustafa Safdari, Clément Crepy, Luning Sun, Stephen Fitz, Peter Romero, Marwa Abdulhai, Aleksandra Faust, Maja Matarić (2025) 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/pmc-12719228.txt.
Description
Gregory Serapio-García and co-authors build a careful way to test, and then steer, the personality that large language models show in their replies. Their 2025 paper notes that chatbots now talk with the public across the world. The human-like traits in their style can shape how well those talks go.
The team adapted standard personality tests for models and checked that the scores hold up. Across eighteen models, the tests proved sound for some models under some prompts, and more so for bigger models that had been tuned to follow commands. The team also showed that a model's style can be pushed toward a chosen trait profile. They close with uses of the method and the ethics of shaping a machine's voice.
Catalog record
| CREATOR | Gregory Serapio-García, Mustafa Safdari, Clément Crepy, Luning Sun, Stephen Fitz, Peter Romero, Marwa Abdulhai, Aleksandra Faust, Maja Matarić |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 84 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |