Towards Auditing Large Language Models: Improving Text-based Stereotype Detection

TL;DR

This paper introduces a new dataset and classifier to detect stereotypes in text, enabling better auditing of large language models for societal biases and demonstrating bias reduction in GPT models over time.

Contribution

The work presents a novel multi-grain stereotype dataset and a classifier, along with experiments showing improved detection and bias assessment in LLMs.

Findings

Multi-class training outperforms binary classification.

The new model uses relevant text features effectively.

Bias in GPT models has decreased over time.

Abstract

Large Language Models (LLM) have made significant advances in the recent past becoming more mainstream in Artificial Intelligence (AI) enabled human-facing applications. However, LLMs often generate stereotypical output inherited from historical data, amplifying societal biases and raising ethical concerns. This work introduces i) the Multi-Grain Stereotype Dataset, which includes 52,751 instances of gender, race, profession and religion stereotypic text and ii) a novel stereotype classifier for English text. We design several experiments to rigorously test the proposed model trained on the novel dataset. Our experiments show that training the model in a multi-class setting can outperform the one-vs-all binary counterpart. Consistent feature importance signals from different eXplainable AI tools demonstrate that the new model exploits relevant text features. We utilise the newly created model to assess the stereotypic behaviour of the popular GPT family of models and observe the reduction of bias over time. In summary, our work establishes a robust and practical framework for auditing and evaluating the stereotypic bias in LLM.