Agentic data science with the Data Agent Kit
cloud.google.com·7h ago
AI Research
arXiv CS.AI·Aug 20·5 min read
Institutional Prestige as Geographic Bias in Large Language Models: Evidence from Three Factorial Experiments with Bootstrap Confidence Intervals
TL;DR
Large language models (LLMs) exhibit geographic bias, favoring candidates from prestigious institutions over others, regardless of name ethnicity. Experiments reveal that journal prestige significantly outweighs institutional prestige in candidate evaluations.
✦ Why It Matters
Evaluate candidate assessments critically, especially considering institutional and journal prestige biases.
Key Takeaways
- Assess the impact of institutional prestige on candidate evaluations.
- Consider journal prestige when evaluating research contributions.
- Utilize the Neutrosophic Bias Index for deeper bias analysis.
- Be aware of geographic biases in AI evaluations.
How It Works
The study employs factorial experiments to isolate the effects of institutional prestige and geographic location on LLM evaluations. By systematically varying these factors, researchers can quantify biases and their implications on candidate assessments.