New 'Authority Expectancy Effect' identified in LLMs like Claude and Gemini

Researchers have identified a new phenomenon in large language models called the Authority Expectancy Effect (AEE), which describes how social authority signals influence model judgments in multi-user conflict scenarios. This effect was observed across four LLMs—Claude, Gemini, GPT, and Grok—in experiments involving resource allocation, fault attribution, and dispute mediation. The study found that occupational authority, institutional documentation, and relational congruence significantly restructured model judgments beyond simple cue reweighting, demonstrating that AEE is reference-dependent, involves evidential reinterpretation, and exhibits direction sensitivity. AI

IMPACT This research highlights potential biases in LLMs related to social authority, suggesting a need for more nuanced conflict resolution capabilities.

RANK_REASON The cluster contains a research paper published on arXiv detailing a new phenomenon observed in LLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• Authority Expectancy Effect
• CatalyzeX
• Claude
• DagsHub
• Gemini
• generative pre-trained transformer
• Gotit.pub
• Grok
• Hugging Face
• Influence Flower
• ScienceCast

• paper
• safety

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