Marked Personas: Using Natural Language Prompts to Measure Stereotypes in Language Models
Cheng, Myra, Durmus, Esin, Jurafsky, Dan
–arXiv.org Artificial Intelligence
To recognize and mitigate harms from large language models (LLMs), we need to understand the prevalence and nuances of stereotypes in LLM outputs. Toward this end, we present Marked Personas, a prompt-based method to measure stereotypes in LLMs for intersectional demographic groups without any lexicon or data labeling. Grounded in the sociolinguistic concept of markedness (which characterizes explicitly linguistically marked categories versus unmarked defaults), our proposed method is twofold: 1) prompting an LLM to generate personas, i.e., natural language descriptions, of the target demographic group alongside personas of unmarked, default groups; 2) identifying the words that significantly distinguish personas of the target group from corresponding unmarked ones. We find that the portrayals generated by GPT-3.5 and GPT-4 contain higher rates of racial stereotypes than human-written portrayals using the same prompts. The words distinguishing personas of marked (non-white, non-male) groups reflect patterns of othering and exoticizing these demographics. An intersectional lens further reveals tropes that dominate portrayals of marginalized groups, such as tropicalism and the hypersexualization of minoritized women. These representational harms have concerning implications for downstream applications like story generation.
arXiv.org Artificial Intelligence
May-29-2023
- Country:
- Africa > Middle East (0.04)
- North America
- Mexico (0.04)
- United States
- Texas (0.04)
- Minnesota (0.04)
- New York
- Richmond County > New York City (0.04)
- Queens County > New York City (0.04)
- New York County > New York City (0.04)
- Kings County > New York City (0.04)
- Bronx County > New York City (0.04)
- California > Santa Clara County
- Palo Alto (0.04)
- Europe
- Spain (0.04)
- Middle East (0.04)
- Croatia (0.04)
- United Kingdom > England
- Oxfordshire > Oxford (0.04)
- Sweden > Stockholm
- Stockholm (0.04)
- Asia
- Genre:
- Research Report > New Finding (0.46)
- Industry:
- Law (0.93)
- Law Enforcement & Public Safety > Crime Prevention & Enforcement (0.92)
- Leisure & Entertainment (0.67)
- Health & Medicine > Therapeutic Area (0.67)
- Technology: