“Turning right”? An experimental study on the political value shift in large language models
Guardado en:
| Udgivet i: | Humanities & Social Sciences Communications vol. 12, no. 1 (Dec 2025), p. 179 |
|---|---|
| Udgivet: |
Springer Nature B.V.
|
| Fag: | |
| Online adgang: | Citation/Abstract Full Text Full Text - PDF |
| Tags: |
Ingen Tags, Vær først til at tagge denne postø!
|
MARC
| LEADER | 00000nab a2200000uu 4500 | ||
|---|---|---|---|
| 001 | 3165258100 | ||
| 003 | UK-CbPIL | ||
| 022 | |a 2662-9992 | ||
| 022 | |a 2055-1045 | ||
| 024 | 7 | |a 10.1057/s41599-025-04465-z |2 doi | |
| 035 | |a 3165258100 | ||
| 045 | 2 | |b d20251201 |b d20251231 | |
| 245 | 1 | |a “Turning right”? An experimental study on the political value shift in large language models | |
| 260 | |b Springer Nature B.V. |c Dec 2025 | ||
| 513 | |a Journal Article | ||
| 520 | 3 | |a Constructing artificial intelligence that aligns with human values is a crucial challenge, with political values playing a distinctive role among various human value systems. In this study, we adapted the Political Compass Test and combined it with rigorous bootstrapping techniques to create a standardized method for testing political values in AI. This approach was applied to multiple versions of ChatGPT, utilizing a dataset of over 3000 tests to ensure robustness. Our findings reveal that while newer versions of ChatGPT consistently maintain values within the libertarian-left quadrant, there is a statistically significant rightward shift in political values over time, a phenomenon we term a ‘value shift’ in large language models. This shift is particularly noteworthy given the widespread use of LLMs and their potential influence on societal values. Importantly, our study controlled for factors such as user interaction and language, and the observed shifts were not directly linked to changes in training datasets. While this research provides valuable insights into the dynamic nature of value alignment in AI, it also underscores limitations, including the challenge of isolating all external variables that may contribute to these shifts. These findings suggest a need for continuous monitoring of AI systems to ensure ethical value alignment, particularly as they increasingly integrate into human decision-making and knowledge systems. | |
| 610 | 4 | |a OpenAI | |
| 653 | |a Data analysis | ||
| 653 | |a Data collection | ||
| 653 | |a Research methodology | ||
| 653 | |a Algorithms | ||
| 653 | |a Ideology | ||
| 653 | |a Experiments | ||
| 653 | |a Large language models | ||
| 653 | |a Chatbots | ||
| 653 | |a Questionnaires | ||
| 653 | |a Political parties | ||
| 653 | |a Social norms | ||
| 653 | |a Bias | ||
| 653 | |a Ethics | ||
| 653 | |a Bootstrapping | ||
| 653 | |a Politics | ||
| 653 | |a Robustness | ||
| 653 | |a Libertarianism | ||
| 653 | |a Models | ||
| 653 | |a Values | ||
| 653 | |a Decision making | ||
| 653 | |a Artificial intelligence | ||
| 653 | |a Alignment | ||
| 653 | |a Humans | ||
| 653 | |a Human-computer interaction | ||
| 653 | |a Language attitudes | ||
| 653 | |a Language shift | ||
| 653 | |a Bootstrap method | ||
| 653 | |a Language modeling | ||
| 653 | |a Language | ||
| 653 | |a Tests | ||
| 773 | 0 | |t Humanities & Social Sciences Communications |g vol. 12, no. 1 (Dec 2025), p. 179 | |
| 786 | 0 | |d ProQuest |t Social Science Database | |
| 856 | 4 | 1 | |3 Citation/Abstract |u https://www.proquest.com/docview/3165258100/abstract/embedded/7BTGNMKEMPT1V9Z2?source=fedsrch |
| 856 | 4 | 0 | |3 Full Text |u https://www.proquest.com/docview/3165258100/fulltext/embedded/7BTGNMKEMPT1V9Z2?source=fedsrch |
| 856 | 4 | 0 | |3 Full Text - PDF |u https://www.proquest.com/docview/3165258100/fulltextPDF/embedded/7BTGNMKEMPT1V9Z2?source=fedsrch |