Beyond Static Personas: Situational Personality Steering for Large Language Models
For developers of personalized LLMs, this work enables more adaptable and controllable personality modeling without resource-intensive training, addressing a key bottleneck in human-centric AI interactions.
The paper addresses the limitations of static personality modeling in LLMs by proposing IRIS, a training-free framework for situational personality steering. IRIS achieves superior performance over baselines on PersonalityBench and a new benchmark SPBench, demonstrating generalization and robustness across unseen situations and model architectures.
Personalized Large Language Models (LLMs) facilitate more natural, human-like interactions in human-centric applications. However, existing personalization methods are constrained by limited controllability and high resource demands. Furthermore, their reliance on static personality modeling restricts adaptability across varying situations. To address these limitations, we first demonstrate the existence of situation-dependency and consistent situation-behavior patterns within LLM personalities through a multi-perspective analysis of persona neurons. Building on these insights, we propose IRIS, a training-free, neuron-based Identify-Retrieve-Steer framework for advanced situational personality steering. Our approach comprises situational persona neuron identification, situation-aware neuron retrieval, and similarity-weighted steering. We empirically validate our framework on PersonalityBench and our newly introduced SPBench, a comprehensive situational personality benchmark. Experimental results show that our method surpasses best-performing baselines, demonstrating IRIS's generalization and robustness to complex, unseen situations and different models architecture.