Can LLM be a Personalized Judge?

2024/06/17 by Yijiang River Dong, Tiancheng Hu, Dong, Yijiang River +3 · 47 citations
Psychology · Social Sciences · #Computation and Language (cs.CL) #Computer science #Computers and Society (cs.CY) #FOS: Computer and information sciences #Judicial and Constitutional Studies #Legal Education and Practice Innovations #Legal Systems and Judicial Processes #Psychology

paper · pdf · doi:10.48550/arxiv.2406.11657

published in arXiv (Cornell University) (Cornell University)

openalex publication_date 2024/06/17 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/01

Abstract

Ensuring that large language models (LLMs) reflect diverse user values and preferences is crucial as their user bases expand globally. It is therefore encouraging to see the growing interest in LLM personalization within the research community. However, current works often rely on the LLM-as-a-Judge approach for evaluation without thoroughly examining its validity. In this paper, we investigate the reliability of LLM-as-a-Personalized-Judge, asking LLMs to judge user preferences based on personas. Our findings suggest that directly applying LLM-as-a-Personalized-Judge is less reliable than previously assumed, showing low and inconsistent agreement with human ground truth. The personas typically used are often overly simplistic, resulting in low predictive power. To address these issues, we introduce verbal uncertainty estimation into the LLM-as-a-Personalized-Judge pipeline, allowing the model to express low confidence on uncertain judgments. This adjustment leads to much higher agreement (above 80%) on high-certainty samples for binary tasks. Through human evaluation, we find that the LLM-as-a-Personalized-Judge achieves comparable performance to third-party humans evaluation and even surpasses human performance on high-certainty samples. Our work indicates that certainty-enhanced LLM-as-a-Personalized-Judge offers a promising direction for developing more reliable and scalable methods for evaluating LLM personalization.

Cited by

• Informing AI Policy Assessment using Large-Scale Simulation of Interventions
• From Personalization to Prejudice: Bias and Discrimination in Memory-Enhanced AI Agents for Recruitment
• Few-Shot Inference of Human Perceptions of Robot Performance in Social Navigation Scenarios
• CrowdLLM: Building LLM-Based Digital Populations Augmented with Generative Models
• Generate, Evaluate, Iterate: Synthetic Data for Human-in-the-Loop Refinement of LLM Judges
• Towards Aligning Multimodal LLMs with Human Experts: A Focus on Parent-Child Interaction
• Deciphering Personalization: Towards Fine-Grained Explainability in Natural Language for Personalized Image Generation Models
• Instant Personalized Large Language Model Adaptation via Hypernetwork
• SimBench: Benchmarking the Ability of Large Language Models to Simulate Human Behaviors
• Confidence as a Reward: Transforming LLMs into Reward Models
• Taking a SEAT: Predicting Value Interpretations from Sentiment, Emotion, Argument, and Topic Annotations
• Flipping the Dialogue: Training and Evaluating User Language Models
• Reasoning Beyond Majority Vote: An Explainable SpeechLM Framework for Speech Emotion Recognition
• Quantifying the Impact of Structured Output Format on Large Language Models through Causal Inference
• Doubly-Robust LLM-as-a-Judge: Externally Valid Estimation with Imperfect Personas
• Comparative Personalization for Multi-document Summarization
• When Avatars Have Personality: Effects on Engagement and Communication in Immersive Medical Training
• Online Anti-sexist Speech: Identifying Resistance to Gender Bias in Political Discourse
• LaajMeter: A Framework for LaaJ Evaluation
• Evaluating Podcast Recommendations with Profile-Aware LLM-as-a-Judge
• Can You Trick the Grader? Adversarial Persuasion of LLM Judges
• Improved Personalized Headline Generation via Denoising Fake Interests from Implicit Feedback
• Beyond the Surface: Enhancing LLM-as-a-Judge Alignment with Human via Internal Representations
• TRIDENT: Benchmarking LLM Safety in Finance, Medicine, and Law
• Anthropomimetic Uncertainty: What Verbalized Uncertainty in Language Models is Missing
• Hateful Person or Hateful Model? Investigating the Role of Personas in Hate Speech Detection by Large Language Models
• Measuring AI Alignment with Human Flourishing
• PRIME: Large Language Model Personalization with Cognitive Dual-Memory and Personalized Thought Process
• Improving the Distributional Alignment of LLMs using Supervision
• Efficient Online RFT with Plug-and-Play LLM Judges: Unlocking State-of-the-Art Performance
• SynthesizeMe! Inducing Persona-Guided Prompts for Personalized Reward Models in LLMs
• CETBench: A Novel Dataset constructed via Transformations over Programs for Benchmarking LLMs for Code-Equivalence Checking
• Localizing Persona Representations in LLMs
• MEDAL: A Framework for Benchmarking LLMs as Multilingual Open-Domain Dialogue Evaluators
• Amulet: Putting Complex Multi-Turn Conversations on the Stand with LLM Juries
• Is Active Persona Inference Necessary for Aligning Small Models to Personal Preferences?
• MedGUIDE: Benchmarking Clinical Decision-Making in Large Language Models
• Re-Centering Humans in LLM Personalization
• Evaluating Second-Order Bias of LLMs Through Epistemic Entitlement
• LLM-Evaluation Tropes: Perspectives on the Validity of LLM-Evaluations
• Toward Safe and Human-Aligned Game Conversational Recommendation via Multi-Agent Decomposition
• Toward Generalizable Evaluation in the LLM Era: A Survey Beyond Benchmarks
• LoRe: Personalizing LLMs via Low-Rank Reward Modeling
• Know Me, Respond to Me: Benchmarking LLMs for Dynamic User Profiling and Personalized Responses at Scale
• Persona-judge: Personalized Alignment of Large Language Models via Token-level Self-judgment
• A Survey of Personalization: From RAG to Agent
• How Is Generative AI Used for Persona Development?: A Systematic Review of 52 Research Articles

Related

• Re-Centering Humans in LLM Personalization
• Aligning Black-box Language Models with Human Judgments
• A Survey on LLM-as-a-Judge
• Judging the Judges: Evaluating Alignment and Vulnerabilities in LLMs-as-Judges
• Human-Centered Design Recommendations for LLM-as-a-Judge