LLMs display left bias on abstract policy questions but align with centrist parties and exhibit change-aversion on real Swiss federal referenda.
The political biases of ChatGPT.Social Sciences, 12(3):148
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
years
2026 5verdicts
UNVERDICTED 5representative citing papers
MÖVE presents a new German-language benchmark evaluating 39 LLMs on performance and governance criteria using ten public-administration datasets.
LLMs exhibit higher perplexity on far-right and nationalist party texts than social-democratic ones, consistent across models and languages with correlation to translation metrics.
Centralized matching mechanisms outperform free negotiation in stability and efficiency with LLM agents, who also report preferences truthfully more often than humans, though not always in line with strategy-proofness predictions.
PCT is a reinforcement learning approach that trains LLMs for symmetric sentiment and helpfulness across paired opposing political prompts, reducing covert bias while preserving general performance.
citing papers explorer
-
M\"OVE: A Holistic LLM Benchmark for the German Public Sector
MÖVE presents a new German-language benchmark evaluating 39 LLMs on performance and governance criteria using ten public-administration datasets.
-
Large Language Models are Perplexed by some Political Parties
LLMs exhibit higher perplexity on far-right and nationalist party texts than social-democratic ones, consistent across models and languages with correlation to translation metrics.
-
Reducing Political Manipulation with Consistency Training
PCT is a reinforcement learning approach that trains LLMs for symmetric sentiment and helpfulness across paired opposing political prompts, reducing covert bias while preserving general performance.