Expert Personas Improve LLM Alignment but Damage Accuracy: Bootstrapping Intent-Based Persona Routing with PRISM

• URL: http://arxiv.org/abs/2603.18507v1
• Date: Thu, 19 Mar 2026 05:28:21 GMT
• Title: Expert Personas Improve LLM Alignment but Damage Accuracy: Bootstrapping Intent-Based Persona Routing with PRISM
• Abstract summary: We study how model optimization, task type, prompt length, and placement can impact expert persona effectiveness across instruction-tuned and reasoning LLMs.<n>Based on our findings, we developed a pipeline to fully leverage the benefits of an expert persona, named PRISM.
• Score: 21.495137998180954
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Persona prompting can steer LLM generation towards a domain-specific tone and pattern. This behavior enables use cases in multi-agent systems where diverse interactions are crucial and human-centered tasks require high-level human alignment. Prior works provide mixed opinions on their utility: some report performance gains when using expert personas for certain domains and their contribution to data diversity in synthetic data creation, while others find near-zero or negative impact on general utility. To fully leverage the benefits of the LLM persona and avoid its harmfulness, a more comprehensive investigation of the mechanism is crucial. In this work, we study how model optimization, task type, prompt length, and placement can impact expert persona effectiveness across instruction-tuned and reasoning LLMs, and provide insight into conditions under which expert personas fail and succeed. Based on our findings, we developed a pipeline to fully leverage the benefits of an expert persona, named PRISM (Persona Routing via Intent-based Self-Modeling), which self-distills an intent-conditioned expert persona into a gated LoRA adapter through a bootstrapping process that requires no external data, models, or knowledge. PRISM enhances human preference and safety alignment on generative tasks while maintaining accuracy on discriminative tasks across all models, with minimal memory and computing overhead.

Related papers

• Hierarchical Multi-Persona Induction from User Behavioral Logs: Learning Evidence-Grounded and Truthful Personas [7.672918013974333]
  We propose a hierarchical framework that aggregates user actions into intent memories and induces multiple evidence-grounded personas.<n>Experiments on a large-scale service log and two public datasets show that our method induces more coherent, evidence-grounded, and trustworthy personas.
  arXiv  Detail & Related papers  (2026-04-28T21:13:16Z)
• How Does Personalized Memory Shape LLM Behavior? Benchmarking Rational Preference Utilization in Personalized Assistants [25.83552447206606]
  Large language model (LLM)-powered assistants have recently integrated memory mechanisms that record user preferences, leading to more personalized and user-aligned responses.<n>We develop RPEval, a benchmark comprising a personalized intent reasoning dataset and a multi-granularity evaluation protocol.<n>RPEval reveals the widespread phenomenon of irrational personalization in existing LLMs and, through error pattern analysis, illustrates its negative impact on user experience.<n>We introduce RP-Reasoner, which treats memory utilization as a pragmatic reasoning process, enabling the selective integration of personalized information.
  arXiv  Detail & Related papers  (2026-01-23T10:19:48Z)
• HumanLLM: Towards Personalized Understanding and Simulation of Human Nature [72.55730315685837]
  HumanLLM is a foundation model designed for personalized understanding and simulation of individuals.<n>We first construct the Cognitive Genome, a large-scale corpus curated from real-world user data on platforms like Reddit, Twitter, Blogger, and Amazon.<n>We then formulate diverse learning tasks and perform supervised fine-tuning to empower the model to predict a wide range of individualized human behaviors, thoughts, and experiences.
  arXiv  Detail & Related papers  (2026-01-22T09:27:27Z)
• Learning Individual Intrinsic Reward in Multi-Agent Reinforcement Learning via Incorporating Generalized Human Expertise [6.441011477647557]
  Efficient exploration in multi-agent reinforcement learning (MARL) is a challenging problem when receiving only a team reward.<n>A powerful method to mitigate this issue involves crafting dense individual rewards to guide the agents toward efficient exploration.<n>We propose a novel framework, LIGHT, which can integrate human knowledge into MARL algorithms in an end-to-end manner.
  arXiv  Detail & Related papers  (2025-07-25T00:59:10Z)
• PersonaAgent: When Large Language Model Agents Meet Personalization at Test Time [87.99027488664282]
  PersonaAgent is a framework designed to address versatile personalization tasks.<n>It integrates a personalized memory module and a personalized action module.<n>Test-time user-preference alignment strategy ensures real-time user preference alignment.
  arXiv  Detail & Related papers  (2025-06-06T17:29:49Z)
• OPeRA: A Dataset of Observation, Persona, Rationale, and Action for Evaluating LLMs on Human Online Shopping Behavior Simulation [56.47029531207105]
  OPERA is the first public dataset that comprehensively captures user personas, browser observations, fine-grained web actions, and self-reported just-in-time rationales.<n>We establish the first benchmark to evaluate how well current LLMs can predict a specific user's next action and rationale.
  arXiv  Detail & Related papers  (2025-06-05T21:37:49Z)
• AgentDAM: Privacy Leakage Evaluation for Autonomous Web Agents [66.29263282311258]
  We introduce a new benchmark AgentDAM that measures if AI web-navigation agents follow the privacy principle of data minimization''<n>Our benchmark simulates realistic web interaction scenarios end-to-end and is adaptable to all existing web navigation agents.
  arXiv  Detail & Related papers  (2025-03-12T19:30:31Z)
• Forewarned is Forearmed: Leveraging LLMs for Data Synthesis through Failure-Inducing Exploration [90.41908331897639]
  Large language models (LLMs) have significantly benefited from training on diverse, high-quality task-specific data. We present a novel approach, ReverseGen, designed to automatically generate effective training samples.
  arXiv  Detail & Related papers  (2024-10-22T06:43:28Z)
• PersLLM: A Personified Training Approach for Large Language Models [66.16513246245401]
  We propose PersLLM, a framework for better data construction and model tuning.<n>For insufficient data usage, we incorporate strategies such as Chain-of-Thought prompting and anti-induction.<n>For rigid behavior patterns, we design the tuning process and introduce automated DPO to enhance the specificity and dynamism of the models' personalities.
  arXiv  Detail & Related papers  (2024-07-17T08:13:22Z)
• Task-aware Privacy Preservation for Multi-dimensional Data [4.138783926370621]
  Local differential privacy (LDP) is a state-of-the-art technique for privacy preservation. In the future, LDP can be adopted to anonymize richer user data attributes. We show how to significantly improve the ultimate task performance for multi-dimensional user data by considering a task-aware privacy preservation problem.
  arXiv  Detail & Related papers  (2021-10-05T20:03:53Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.