🤖 AI Summary
This study addresses the limitations of existing approaches that rely solely on unidimensional sentiment polarity to assess large language models’ (LLMs) attitudes toward political and geopolitical entities, which fails to capture the nuanced complexity of regional affective expression. Introducing the valence–arousal–dominance (VAD) three-dimensional affect model to this domain for the first time, we systematically analyze affective disparities across 19 LLMs toward 500 post-Soviet targets. Leveraging large-scale model querying, dual-LLM automated scoring (GPT-4o-mini and Qwen3.6-35B-A3B), human annotation validation, and Ward hierarchical clustering, we identify three distinct behavioral clusters transcending model origin, architecture, and scale. Our analysis further reveals a strong negative correlation (r = –0.81) between generic response tendencies and arousal levels, demonstrating that the VAD framework effectively captures affective intensity and dominance dimensions overlooked by conventional methods, thereby enabling deeper structural insights into LLM emotional expression.
📝 Abstract
Large language models trained and aligned within different linguistic and regional ecosystems may frame the same political, cultural, and geopolitical entities in different ways. Such differences are often evaluated through sentiment, favorability, or stance, reducing model attitudes to a single positive-negative axis. We introduce REGARD, a study of what drives affective framing differences across LLMs on post-Soviet entities using target-directed Valence-Arousal-Dominance profiling. We query 19 models on 500 region-specific targets, score their responses with two independent LLM judges, GPT-4o-mini and Qwen3.6-35B-A3B, and validate the measurements on a 300-item human-annotated subset. Post-hoc Ward-linkage clustering of all 19 models by affective and response-behavior profiles yields three behavioral clusters that cut across model origin, family, and parameter count. Generic-answer rate is strongly associated with lower arousal (r = -0.81) and with cluster placement: models that deflect evaluative prompts with templated responses cluster together at low arousal regardless of origin. These findings show that VAD profiling captures emotional intensity, a dimension of affective framing that is largely invisible to conventional sentiment-based evaluation.