Group Alignment-Induced Sycophancy: A Two-Sided Evaluation of Steerable Pluralistic Alignment

2026-08-11

AI Summary

This study addresses a critical oversight in existing group alignment methods, which prioritize consistency between models and group viewpoints while neglecting the resulting sycophantic behavior. Introducing the novel concept of “Group Alignment–induced Sycophancy” (GAS), this work proposes a dual-aspect, multidimensional evaluation framework to systematically assess three alignment approaches across four large language models and thirteen distinct demographic groups. Through multi-model, multi-group comparative experiments and quantitative metric analysis, the study reveals that alignment gains and sycophancy shifts exhibit significant non-uniform distributions across groups. These findings underscore the necessity of replacing single scalar alignment scores with group-specific, multidimensional profiles to enable fairer and more fine-grained alignment evaluation.

Abstract

Group alignment adapts a language model to a demographic group to produce responses that reflect the group's opinions, values, and preferences. Sycophancy, a well-documented by-product of alignment, causes the model to over-agree with the user regardless of factual and objective information. However, existing group alignment methods and evaluations focus only on how closely the model matches the group's opinions, overlooking the induced change in sycophantic behaviour. To bridge this gap, we introduce \textbf{G}roup \textbf{A}lignment-induced \textbf{S}ycophancy (GAS) and systematically evaluate alignment across 3 methods, 4 models and 13 demographic groups, on both the intended gain in opinion alignment and the unintended shift in sycophancy. We find that gain and shift are non-uniform across groups: under an identical budget, some groups receive larger gains in opinion alignment than others, and the induced sycophancy shift forms a group-specific profile rather than a single-dimensional change. These results suggest that group alignment should be reported as a two-sided, multi-dimensional profile rather than a single fit score that accounts for per-group differences when adapting LLMs to diverse populations.

Problem

• group alignment
• sycophancy
• language models
• demographic groups
• alignment evaluation

Innovation

• Group Alignment
• Sycophancy
• Pluralistic Alignment
• Demographic Bias
• Language Model Evaluation