Problem
Research questions and friction points this paper is trying to address.
Generative Monoculture
Topic Saliency Shift
Large Language Models
Homogenisation
Perspective Diversity
Narrowing the Horizon: Quantifying Topic Saliency Shifts in Generative Monoculture
🤖 AI Summary
This study addresses the risk of cultural homogenization arising from output convergence in large language models, where marginalized topics remain difficult to identify. We propose a fine-grained topic salience measurement framework that integrates topic modeling, cross-model comparison, and intervention testing to quantitatively analyze topic evolution and perspective shifts across different model families during post-training. Our analysis reveals discourse homogenization on issues such as climate change and demonstrates the effectiveness of specialized models in preserving diverse viewpoints. Ultimately, this work provides critical monitoring and intervention mechanisms for assessing risks to AI knowledge diversity and safeguarding intellectual pluralism.
📝 Abstract
As Large Language Models (LLMs) become central to how we access and share information, they play an increasingly powerful role in shaping global knowledge. However, as these models evolve, their outputs risk converging into a \textit{generative monoculture}, where the diversity of perspectives they represent narrows over time. Studies at the model level often fail to pinpoint which specific topics or viewpoints are being marginalised or amplified in this process. In this paper, we introduce a method to measure shifts in topic saliency across model families, tracking what gains or loses prominence during post-training. Applying this approach to a case study of climate change discourse, we demonstrate how homogenisation affects the representation of diverse solutions across different models. We also test interventions to counter this trend, showing that specialised models can help preserve a broader range of perspectives. This underscores the importance of monitoring topic saliency to diagnose the risks of monoculture and to ensure AI systems reflect a pluralism of ideas. Data and Code are accessible \href{https://github.com/oriane/topic_saliency_shift}{here}.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Topic Saliency Shifts
Generative Monoculture
Large Language Models
Homogenisation
Specialised Models
🔎 Similar Papers
No similar papers found.
