Ideological Bias in LLMs' Economic Causal Reasoning

arXiv:2604.21334v1 Announce Type: new Abstract: Do large language models (LLMs) exhibit systematic ideological bias when reasoning about economic causal effects? As LLMs are increasingly used in policy analysis and economic reporting, where directionally correct causal judgments are essential, this...