Racial Groups’ Metastereotypes about Artificial Intelligence Large Language Models

In psychology, stereotypes are defined as descriptive or evaluative characteristics associated with social categories (e.g., “doctors are smart”). Metastereotypes are beliefs that a person holds about how their group is stereotyped by others. Despite growing knowledge of how Artificial Intelligence Large Language Models (LLMs) reproduce stereotypes, little is known about how members of various racial groups anticipate being stereotyped by LLMs, whether those beliefs are accurate, and how they shape engagement with AI systems. Psychologist Gandalf Nicolas will examine racial groups’ metastereotypes about LLMs. He will conduct large scale text analysis and surveys for his study.