The Homogenization Problem in LLMs: Towards Meaningful Diversity in AI Safety

TL;DR

This paper addresses homogenization in Large Language Models, proposing a framework to measure and mitigate bias and promote diversity, with implications for AI safety and social equity.

Contribution

It introduces a novel framework for characterizing homogenization in LLMs, formalizes it using queer and feminist theory, and demonstrates its application through bias detection.

Findings

Identified gender bias in Claude 3.5 Haiku on an open-ended prompt.

Proposed a formal framework to encode stakeholder values regarding diversity.

Introduced the concept of xeno-reproduction to promote diversity in AI tasks.

Abstract

Generative AI models reproduce the human biases in their training data and further amplify them through mechanisms such as mode collapse. The loss of diversity produces homogenization, which not only harms the minoritized but impoverishes everyone. We argue homogenization should be a central concern in AI safety. To meaningfully characterize homogenization in Large Language Models (LLMs), we introduce a framework that allows stakeholders to encode their context and value system. We illustrate our approach with an experiment that surfaces gender bias in an LLM (Claude 3.5 Haiku) on an open-ended story prompt. Building from queer theory, we formalize homogenization in terms of normativity. Borrowing language from feminist theory, we introduce the concept of xeno-reproduction as a class of tasks for mitigating homogenization by promoting diversity. Our work opens a collaborative line of research that seeks to understand and advance diversity in AI.