PersonaEval: Are LLM Evaluators Human Enough to Judge Role-Play?

• URL: http://arxiv.org/abs/2508.10014v1
• Date: Wed, 06 Aug 2025 13:06:15 GMT
• Title: PersonaEval: Are LLM Evaluators Human Enough to Judge Role-Play?
• Authors: Lingfeng Zhou, Jialing Zhang, Jin Gao, Mohan Jiang, Dequan Wang,
• Abstract summary: We present PersonaEval, the first benchmark designed to test whether role evaluators can reliably identify human roles.<n>Our experiments, including a human study, show that even the best-performing LLMs reach only around 69% accuracy.<n>To better understand this gap, we examine training-time adaptation and test-time compute, suggesting that reliable evaluation requires more than task-specific tuning.
• Score: 13.537861623851054
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Current role-play studies often rely on unvalidated LLM-as-a-judge paradigms, which may fail to reflect how humans perceive role fidelity. A key prerequisite for human-aligned evaluation is role identification, the ability to recognize who is speaking based on dialogue context. We argue that any meaningful judgment of role-playing quality (how well a character is played) fundamentally depends on first correctly attributing words and actions to the correct persona (who is speaking). We present PersonaEval, the first benchmark designed to test whether LLM evaluators can reliably identify human roles. PersonaEval uses human-authored dialogues from novels, scripts, and video transcripts, challenging models to determine the correct persona according to the conversation context. Our experiments, including a human study, show that even the best-performing LLMs reach only around 69% accuracy, well below the level needed for reliable evaluation. In contrast, human participants perform near ceiling with 90.8% accuracy, highlighting that current LLM evaluators are still not human enough to effectively judge role-play scenarios. To better understand this gap, we examine training-time adaptation and test-time compute, suggesting that reliable evaluation requires more than task-specific tuning, but depends on strong, human-like reasoning abilities in LLM evaluators. We release our benchmark at https://github.com/maple-zhou/PersonaEval.

Related papers

• On Evaluating LLM Alignment by Evaluating LLMs as Judges [68.15541137648721]
  evaluating large language models' (LLMs) alignment requires them to be helpful, honest, safe, and to precisely follow human instructions.<n>We examine the relationship between LLMs' generation and evaluation capabilities in aligning with human preferences.<n>We propose a benchmark that assesses alignment without directly evaluating model outputs.
  arXiv  Detail & Related papers  (2025-11-25T18:33:24Z)
• Evaluating LLM-Generated Versus Human-Authored Responses in Role-Play Dialogues [4.4260965078852275]
  This study compares LLM-generated and human-authored responses in multi-turn professional training simulations.<n>Human evaluation revealed significant degradation in LLM-generated response quality across turns.<n>Participants also indicated a consistent preference for human-authored dialogue.
  arXiv  Detail & Related papers  (2025-09-22T12:33:02Z)
• Towards Understanding the Robustness of LLM-based Evaluations under Perturbations [9.944512689015998]
  Large Language Models (LLMs) can serve as automatic evaluators for non-standardized metrics in summarization and dialog-based tasks.<n>We conduct experiments across multiple prompting strategies to examine how LLMs fare as quality evaluators when compared with human judgments.
  arXiv  Detail & Related papers  (2024-12-12T13:31:58Z)
• Thinking Before Speaking: A Role-playing Model with Mindset [0.6428333375712125]
  Large Language Models (LLMs) are skilled at simulating human behaviors. These models tend to perform poorly when confronted with knowledge that the assumed role does not possess. We propose a Thinking Before Speaking (TBS) model in this paper.
  arXiv  Detail & Related papers  (2024-09-14T02:41:48Z)
• Evaluating the Evaluator: Measuring LLMs' Adherence to Task Evaluation Instructions [18.93335792080899]
  We investigate how much influence prompting the LLMs-as-a-judge has on the alignment of AI judgements to human judgements. We aggregate a taxonomy of quality criteria commonly used across state-of-the-art evaluations with LLMs and provide this as a rigorous benchmark of models as judges.
  arXiv  Detail & Related papers  (2024-08-16T14:49:35Z)
• Rel-A.I.: An Interaction-Centered Approach To Measuring Human-LM Reliance [73.19687314438133]
  We study how reliance is affected by contextual features of an interaction. We find that contextual characteristics significantly affect human reliance behavior. Our results show that calibration and language quality alone are insufficient in evaluating the risks of human-LM interactions.
  arXiv  Detail & Related papers  (2024-07-10T18:00:05Z)
• Auto-Arena: Automating LLM Evaluations with Agent Peer Battles and Committee Discussions [77.66677127535222]
  Auto-Arena is an innovative framework that automates the entire evaluation process using LLM-powered agents. In our experiments, Auto-Arena shows a 92.14% correlation with human preferences, surpassing all previous expert-annotated benchmarks.
  arXiv  Detail & Related papers  (2024-05-30T17:19:19Z)
• Character is Destiny: Can Role-Playing Language Agents Make Persona-Driven Decisions? [59.0123596591807]
  We benchmark the ability of Large Language Models (LLMs) in persona-driven decision-making. We investigate whether LLMs can predict characters' decisions provided by the preceding stories in high-quality novels. The results demonstrate that state-of-the-art LLMs exhibit promising capabilities in this task, yet substantial room for improvement remains.
  arXiv  Detail & Related papers  (2024-04-18T12:40:59Z)
• Exploring the Reliability of Large Language Models as Customized Evaluators for Diverse NLP Tasks [65.69651759036535]
  We analyze whether large language models (LLMs) can serve as reliable alternatives to humans.<n>This paper explores both conventional tasks (e.g., story generation) and alignment tasks (e.g., math reasoning)<n>We find that LLM evaluators can generate unnecessary criteria or omit crucial criteria, resulting in a slight deviation from the experts.
  arXiv  Detail & Related papers  (2023-10-30T17:04:35Z)
• Character-LLM: A Trainable Agent for Role-Playing [67.35139167985008]
  Large language models (LLMs) can be used to serve as agents to simulate human behaviors. We introduce Character-LLM that teach LLMs to act as specific people such as Beethoven, Queen Cleopatra, Julius Caesar, etc.
  arXiv  Detail & Related papers  (2023-10-16T07:58:56Z)
• Can Large Language Models Be an Alternative to Human Evaluations? [80.81532239566992]
  Large language models (LLMs) have demonstrated exceptional performance on unseen tasks when only the task instructions are provided. We show that the result of LLM evaluation is consistent with the results obtained by expert human evaluation.
  arXiv  Detail & Related papers  (2023-05-03T07:28:50Z)

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.