High-Dimension Human Value Representation in Large Language Models
This work addresses the need to assess human value alignment in LLMs before deployment, which is crucial for ensuring ethical AI applications across diverse contexts.
The authors tackled the problem of understanding human value alignment in LLMs by proposing UniVaR, a high-dimensional neural representation of symbolic human value distributions, which they used to visualize and explore value prioritization across 25 languages and cultures in 15 LLMs.
The widespread application of LLMs across various tasks and fields has necessitated the alignment of these models with human values and preferences. Given various approaches of human value alignment, there is an urgent need to understand the scope and nature of human values injected into these LLMs before their deployment and adoption. We propose UniVaR, a high-dimensional neural representation of symbolic human value distributions in LLMs, orthogonal to model architecture and training data. This is a continuous and scalable representation, self-supervised from the value-relevant output of 8 LLMs and evaluated on 15 open-source and commercial LLMs. Through UniVaR, we visualize and explore how LLMs prioritize different values in 25 languages and cultures, shedding light on complex interplay between human values and language modeling.