Uncovering Implicit Gender Bias in Narratives through Commonsense Inference
This addresses the problem of harmful social biases in AI-generated content for users and developers, but is incremental as it extends existing bias analysis to implicit aspects.
The study tackled implicit gender bias in narratives generated by pre-trained language models by using a commonsense reasoning engine to analyze protagonist motivations, attributes, and mental states, finding that female characters are portrayed around appearance and male figures around intellect, aligning with prior explicit bias research.
Pre-trained language models learn socially harmful biases from their training corpora, and may repeat these biases when used for generation. We study gender biases associated with the protagonist in model-generated stories. Such biases may be expressed either explicitly ("women can't park") or implicitly (e.g. an unsolicited male character guides her into a parking space). We focus on implicit biases, and use a commonsense reasoning engine to uncover them. Specifically, we infer and analyze the protagonist's motivations, attributes, mental states, and implications on others. Our findings regarding implicit biases are in line with prior work that studied explicit biases, for example showing that female characters' portrayal is centered around appearance, while male figures' focus on intellect.