mlsanityData to 7 Oct 2026

Self-Preference Bias in LLM-as-a-Judge

arXiv 2410.21819 ↗First posted 29 Oct 2024cs.CLcited by 248

Not on the map. Why?

Cited by 40 in LLM agents and tool use, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Automated evaluation leveraging large language models (LLMs), commonly referred to as LLM evaluators or LLM-as-a-judge, has been widely used in measuring the performance of dialogue systems. However, the self-preference bias in LLMs has posed significant risks, including promoting specific styles or policies intrinsic to the LLMs. Despite the importance of this issue, there is a lack of established methods to measure the self-preference bias quantitatively, and its underlying causes are poorly understood. In this paper, we introduce a novel quantitative metric to measure the self-preference bias. Our experimental results demonstrate that GPT-4 exhibits a significant degree of self-preference bias. To explore the causes, we hypothesize that LLMs may favor outputs that are more familiar to them, as indicated by lower perplexity. We analyze the relationship between LLM evaluations and the perplexities of outputs. Our findings reveal that LLMs assign significantly higher evaluations to outputs with lower perplexity than human evaluators, regardless of whether the outputs were self-generated. This suggests that the essence of the bias lies in perplexity and that the self-preference bias exists because LLMs prefer texts more familiar to them.

Key facts. Cited by 248 papers in the map (data through 7 Oct 2026). Builds on 17 of its 24 references that are in the map. First posted on arXiv 29 Oct 2024.

Builds on

17 of 24 references.

References · 17

  1. 2022Training language models to follow instructions with human feedbackcited by 15,712
  2. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  4. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,317
  5. 2016Equality of Opportunity in Supervised Learningcited by 2,000
  6. 2023Pythia: A Suite for Analyzing Large Language Models Across Training and Scalingcited by 1,734
  7. 2020BLEURT: Learning Robust Metrics for Text Generationcited by 1,196
  8. 2023Teaching Large Language Models to Self-Debugcited by 963
  9. 2024Self-Rewarding Language Modelscited by 654
  10. 2024LLM Evaluators Recognize and Favor Their Own Generationscited by 651
  11. 2021Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLPcited by 336
  12. 2023Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervisioncited by 318
5 more
  1. 2024Large Language Models are Inconsistent and Biased Evaluatorscited by 146
  2. 2024Pride and Prejudice: LLM Amplifies Self-Bias in Self-Refinementcited by 111
  3. 2023Verbosity Bias in Preference Labeling by Large Language Modelscited by 89
  4. 2023Evaluation Metrics in the Era of GPT-4: Reliably Evaluating Large Language Models on Sequence to Sequence Taskscited by 53
  5. 2022On the Limitations of Reference-Free Evaluations of Generated Textcited by 52

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 5; 2025: 68; 2026: 175.

248 papers.

Sort:

By area:

  1. 2024From Generation to Judgment: Opportunities and Challenges of LLM-as-a-judgecited by 480
  2. 2025An Illusion of Progress? Assessing the Current State of Web Agentscited by 165
  3. 2025Preference Leakage: A Contamination Problem in LLM-as-a-judgecited by 119
  4. 2025The FACTS Grounding Leaderboard: Benchmarking LLMs' Ability to Ground Responses to Long-Form Inputcited by 49
  5. 2025ResearchQA: Evaluating Scholarly Question Answering at Scale Across 75 Fields with Survey-Mined Questions and Rubricscited by 44
  6. 2025Correlated Errors in Large Language Modelscited by 42
  7. 2025Evaluating Scoring Bias in LLM-as-a-Judgecited by 42
  8. 2025The Prompt Makes the Person(a): A Systematic Evaluation of Sociodemographic Persona Prompting for Large Language Modelscited by 40
  9. 2025Do LLM Evaluators Prefer Themselves for a Reason?cited by 38
  10. 2025Answer Matching Outperforms Multiple Choice for Language Model Evaluationcited by 37
  11. 2025Play Favorites: A Statistical Method to Measure Self-Bias in LLM-as-a-Judgecited by 35
  12. 2025Beyond the Surface: Measuring Self-Preference in LLM Judgmentscited by 30
38 more
  1. 2025JarvisEvo: Towards a Self-Evolving Photo Editing Agent with Synergistic Editor-Evaluator Optimizationcited by 30
  2. 2025CoSER: A Comprehensive Literary Dataset and Framework for Training and Evaluating LLM Role-Playing and Persona Simulationcited by 26
  3. 2025How Benchmark Prediction from Fewer Data Misses the Markcited by 26
  4. 2025Benchmarking Adversarial Robustness to Bias Elicitation in Large Language Models: Scalable Automated Assessment with LLM-as-a-Judgecited by 23
  5. 2024Evaluating Creative Short Story Generation in Humans and Large Language Modelscited by 21
  6. 2025When Identity Skews Debate: Anonymization for Bias-Reduced Multi-Agent Reasoningcited by 20
  7. 2025Multi-Agent Debate for LLM Judges with Adaptive Stability Detectioncited by 20
  8. 2025Variation in Verification: Understanding Verification Dynamics in Large Language Modelscited by 18
  9. 2025Audio MultiChallenge: A Multi-Turn Evaluation of Spoken Dialogue Systems on Natural Human Interactioncited by 18
  10. 2026Comparison requires valid measurement: Rethinking attack success rate comparisons in AI red teamingcited by 15
  11. 2025The FACTS Leaderboard: A Comprehensive Benchmark for Large Language Model Factualitycited by 14
  12. 2025The Silent Judge: Unacknowledged Shortcut Bias in LLM-as-a-Judgecited by 13
  13. 2024The Road to Artificial SuperIntelligence: A Comprehensive Survey of Superalignmentcited by 11
  14. 2025Towards Multi-dimensional Evaluation of LLM Summarization across Domains and Languagescited by 10
  15. 2025AgentInit: Initializing LLM-based Multi-Agent Systems via Diversity and Expertise Orchestration for Effective and Efficient Collaborationcited by 10
  16. 2025One Battle After Another: Probing LLMs' Limits on Multi-Turn Instruction Following with a Benchmark Evolving Frameworkcited by 9
  17. 2025MindEval: Benchmarking Language Models on Multi-turn Mental Health Supportcited by 9
  18. 2025WaymoQA: A Multi-View Visual Question Answering Dataset for Safety-Critical Reasoning in Autonomous Drivingcited by 9
  19. 2026Self-Preference Bias in Rubric-Based Evaluation of Large Language Modelscited by 9
  20. 2026AeSlides: Incentivizing Aesthetic Layout in LLM-Based Slide Generation via Verifiable Rewardscited by 9
  21. 2026Quantifying and Mitigating Self-Preference Bias of LLM Judgescited by 9
  22. 2026The Fragility Of Moral Judgment In Large Language Modelscited by 8
  23. 2025Beyond Accuracy: The Role of Calibration in Self-Improving Large Language Modelscited by 7
  24. 2025Know Thyself? On the Incapability and Implications of AI Self-Recognitioncited by 7
  25. 2026Paraphrasing Adversarial Attack on LLM-as-a-Reviewercited by 6
  26. 2026Am I More Pointwise or Pairwise? Revealing Position Bias in Rubric-Based LLM-as-a-Judgecited by 6
  27. 2026Autonomous Research Agents: A Survey of AI Scientists and the Verification Gapcited by 6
  28. 2024SHARP: Unlocking Interactive Hallucination via Stance Transfer in Role-Playing LLMscited by 5
  29. 2025GAEA: A Geolocation Aware Conversational Assistantcited by 5
  30. 2025Retrieval is Not Enough: Enhancing RAG Reasoning through Test-Time Critique and Optimizationcited by 5
  31. 2025Adversarial Attacks on LLM-as-a-Judge Systems: Insights from Prompt Injectionscited by 5
  32. 2025On the Same Wavelength? Evaluating Pragmatic Reasoning in Language Models across Broad Conceptscited by 5
  33. 2026Are LLM Evaluators Really Narcissists? Sanity Checking Self-Preference Evaluationscited by 5
  34. 2026Reward Modeling for Reinforcement Learning-Based LLM Reasoning: Design, Challenges, and Evaluationcited by 5
  35. 2025Dr.Copilot: A Multi-Agent Prompt Optimized Assistant for Improving Patient-Doctor Communication in Romaniancited by 4
  36. 2025RoleConflictBench: A Benchmark of Role Conflict Scenarios for Evaluating LLMs' Contextual Sensitivitycited by 4
  37. 2025Balanced Accuracy: The Right Metric for Evaluating LLM Judges -- Explained through Youden's J statisticcited by 4
  38. 2026PeerRank: Autonomous LLM Evaluation Through Web-Grounded, Bias-Controlled Peer Reviewcited by 4

50 most cited of 248 All 248 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024Dissecting Human and LLM Preferences
  2. 2025Pairwise or Pointwise? Evaluating Feedback Protocols for Bias in LLM-Based Evaluation
  3. 2025Leveraging LLMs as Meta-Judges: A Multi-Agent Framework for Evaluating LLM Judgments
  4. 2026DIAL: Position-Debiased LLM Judges with Adaptive Human Preference Calibration
  5. 2026Automated Concept Discovery for LLM-as-a-Judge Preference Analysis
  6. 2024Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference
  7. 2026A Consensus-Based Framework for Relative Preference Evaluation of Large Language Models
  8. 2026Toward Robust LLM-Based Judges: Taxonomic Bias Evaluation and Debiasing Optimization