Study finds LLMs show ideological bias in economic reasoning

A new study published on arXiv evaluates whether large language models (LLMs) exhibit ideological bias in their economic causal reasoning. Researchers found that LLMs are less accurate on ideologically contested economic questions, with a systematic bias favoring intervention-oriented perspectives over market-oriented ones. This bias persists even with one-shot prompting, suggesting a need for direction-aware evaluations in policy and economic analysis. AI

IMPACT Highlights potential biases in LLMs used for economic policy analysis, underscoring the need for careful evaluation.

RANK_REASON The cluster contains an academic paper detailing a new evaluation of LLM capabilities. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Donggyu Lee
• EconCausal
• Gotit.pub
• Hugging Face
• large language models
• ScienceCast

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →