Persona Bias
Persona bias refers to the skewed or unfair representation of social groups within AI systems, particularly those employing personas in dialogue or narrative generation. Current research focuses on identifying and mitigating these biases, often through the development of new datasets annotated for various bias types (e.g., gender, race, age) and the creation of frameworks for unbiased persona construction. This work is crucial for ensuring fairness and preventing the perpetuation of harmful stereotypes in AI applications, impacting both the ethical development of AI and the accuracy of research relying on self-reported data.
Papers
September 9, 2024
October 8, 2023
June 22, 2022
May 31, 2022