RoleCDE:Benchmarking and Mitigating Role-Alignment Trade-offs in Role-Playing Agents
2026-06-01 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors created RoleCDE, a new test to see how well role-playing AI agents can handle tough choices when their role's values clash with general rules about behaving properly. They made many scenarios where these conflicts happen and found that AI often chooses to follow general rules over the specific role's values, even if told to stick to the role. This tendency stays the same regardless of how hard the dilemma is but changes a lot depending on the role. They also showed that with extra training using RoleCDE, the AI gets better at balancing role values with general rules without losing overall role-playing skills.
Role-playing agentsLarge language modelsRole alignmentValue conflictCognitive dilemmasBenchmarkingFine-tuningDecision makingRole conditioning
Authors
Huayi Lai, Shichao Song, Simin Niu, Hanyu Wang, Jiawei Yang, Zhouxing Wang, Zhiqiang Yin, Xun Liang
Abstract
Role-playing agents(RPAs) are widely used to steer large language models(LLMs) toward role-consistent behavior, yet existing benchmarks mainly evaluate surface-level fidelity and offer limited insight into decision making under role-alignment value conflicts. To address this gap, we introduce RoleCDE, the first benchmark designed to evaluate RPAs under structured conflicts between role-specific values and alignment-oriented constraints. RoleCDE formulates role-aware decision making as cognitive dilemma scenarios, jointly evaluating role-scenario grounding, value conflict resolution, and decision tendencies. The benchmark is constructed at scale, covering approximately 8k diverse role profiles and scenarios and nearly 24k dilemma instances across three difficulty levels and eight role categories. Evaluation of several mainstream LLMs reveals a "Role Value Decoupling" phenomenon, where agents systematically default to alignment-and morality-consistent decisions rather than role-specific values when the two conflict, even under explicit role conditioning. This behavior is largely invariant to dilemma difficulty but varies substantially across role categories. We further show that RoleCDE-based fine-tuning effectively mitigates this decoupling by improving value trade-off reasoning, while preserving general role-playing fidelity and general reasoning performance. Code is available at: https://github.com/rabbitrose/RoleCDE.