Framing the Narrative: Ideological Mimicry in Large Language Models
An arXiv paper that investigates the phenomenon of ideological mimicry in large language models. It proposes the Poli-SHIFT evaluation framework to systematically analyze how user phrasing and political signals cause systematic shifts in LLM political stances.
Timeline (last 90 days)
The paper "Framing the Narrative: Ideological Mimicry in Large Language Models" argues that the political stance of large language models is not fixed, but shifts systematically in response to the way users express themselves.
This paper is currently an arXiv preprint (v1, arXiv:2609.38256v1) and has not yet undergone peer review.
All Facts (2)
The paper "Framing the Narrative: Ideological Mimicry in Large Language Models" argues that the political stance of large language models is not fixed, but shifts systematically in response to the way users express themselves.
50%UnverifiedThis paper is currently an arXiv preprint (v1, arXiv:2609.38256v1) and has not yet undergone peer review.
50%