Intersectional Bias in Causal Language Models
This work addresses the problem of subtle and complex bias in language models for AI ethics and fairness, though it is incremental as it builds on prior tests of auto-regressive models.
The study investigated intersectional bias in GPT-2 and GPT-NEO models by analyzing sentiment in generated sentences from prompts combining up to three social categories, confirming earlier findings of bias in such models and highlighting its resistance to single-category mitigation techniques.
To examine whether intersectional bias can be observed in language generation, we examine \emph{GPT-2} and \emph{GPT-NEO} models, ranging in size from 124 million to ~2.7 billion parameters. We conduct an experiment combining up to three social categories - gender, religion and disability - into unconditional or zero-shot prompts used to generate sentences that are then analysed for sentiment. Our results confirm earlier tests conducted with auto-regressive causal models, including the \emph{GPT} family of models. We also illustrate why bias may be resistant to techniques that target single categories (e.g. gender, religion and race), as it can also manifest, in often subtle ways, in texts prompted by concatenated social categories. To address these difficulties, we suggest technical and community-based approaches need to combine to acknowledge and address complex and intersectional language model bias.