Beyond Static Personas: Situational Personality Steering for Large Language Models

2026-04-15

AI Summary

Existing personalization approaches for large language models rely on static personality modeling, which struggles to adapt to dynamic contexts, offers limited controllability, and incurs high computational overhead. This work proposes IRIS, a novel framework that, for the first time, reveals the context-dependent nature and behavioral consistency of personality-related neurons within language models. IRIS introduces a training-free dynamic control mechanism that enables efficient and controllable context-aware personality generation through contextual personality identification, context-aware neuron retrieval, and similarity-weighted steering. Experimental results demonstrate that IRIS significantly outperforms current methods on both PersonalityBench and a newly introduced SPBench, exhibiting strong generalization capabilities across unseen complex scenarios and diverse model architectures.

Technology Category

• Natural Language Processing: (Large) Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Intelligent Robots: Behavior Learning & Control

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

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.

Problem

• situational personality
• personalized LLMs
• static personas
• controllability
• adaptability

Innovation

• situational personality steering
• persona neurons
• training-free personalization
• neuron-based control
• dynamic persona adaptation