EigenBench: A Comparative Behavioral Measure of Value Alignment
This addresses the problem of measuring subjective value alignment in AI for researchers and developers, though it is incremental as it builds on existing trust algorithms.
The authors tackled the lack of quantitative metrics for value alignment in AI by proposing EigenBench, a black-box method that benchmarks language models' values using an ensemble and a constitution, and showed it aligns closely with human judgments and recovers model rankings on GPQA without ground truth labels.
Aligning AI with human values is a pressing unsolved problem. To address the lack of quantitative metrics for value alignment, we propose EigenBench: a black-box method for comparatively benchmarking language models' values. Given an ensemble of models, a constitution describing a value system, and a dataset of scenarios, our method returns a vector of scores quantifying each model's alignment to the given constitution. To produce these scores, each model judges the outputs of other models across many scenarios, and these judgments are aggregated with EigenTrust (Kamvar et al., 2003), yielding scores that reflect a weighted consensus judgment of the whole ensemble. EigenBench uses no ground truth labels, as it is designed to quantify subjective traits for which reasonable judges may disagree on the correct label. Hence, to validate our method, we collect human judgments on the same ensemble of models and show that EigenBench's judgments align closely with those of human evaluators. We further demonstrate that EigenBench can recover model rankings on the GPQA benchmark without access to objective labels, supporting its viability as a framework for evaluating subjective values for which no ground truths exist.