LLM-Bias Research: Uncovering National Income Bias in Large Language Models' Career Advice

Introduction / Main Floor: LLM-Bias Research: Uncovering National Income Bias in Large Language Models' Career Advice

An empirical study on GPT-4o, Claude, and Gemini found that large language models have systematic biases when generating career advice—they adjust the content of advice based on students' nationality and gender, and students from low-income countries are more likely to be recommended to pursue community service-related careers.