mlsanityData to 7 Oct 2026

When Can LLMs Actually Correct Their Own Mistakes? A Critical Survey of Self-Correction of LLMs

arXiv 2406.01297 ↗First posted 3 Jun 2024cs.CLcited by 230

Not on the map. Why?

Cited by 56 in LLM reasoning, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Self-correction is an approach to improving responses from large language models (LLMs) by refining the responses using LLMs during inference. Prior work has proposed various self-correction frameworks using different sources of feedback, including self-evaluation and external feedback. However, there is still no consensus on the question of when LLMs can correct their own mistakes, as recent studies also report negative results. In this work, we critically survey broad papers and discuss the conditions required for successful self-correction. We first find that prior studies often do not define their research questions in detail and involve impractical frameworks or unfair evaluations that over-evaluate self-correction. To tackle these issues, we categorize research questions in self-correction research and provide a checklist for designing appropriate experiments. Our critical survey based on the newly categorized research questions shows that (1) no prior work demonstrates successful self-correction with feedback from prompted LLMs, except for studies in tasks that are exceptionally suited for self-correction, (2) self-correction works well in tasks that can use reliable external feedback, and (3) large-scale fine-tuning enables self-correction.

Key facts. Cited by 230 papers in the map (data through 7 Oct 2026). Builds on 113 of its 140 references that are in the map. First posted on arXiv 3 Jun 2024.

Builds on

113 of 140 references.

References · 113

  1. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  2. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  3. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,064
  4. 2023Let's Verify Step by Stepcited by 3,598
  5. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,317
  6. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,155
  7. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 2,950
  8. 2023G-Eval: NLG Evaluation using GPT-4 with Better Human Alignmentcited by 2,205
  9. 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflectioncited by 1,800
  10. 2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,739
  11. 2023FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generationcited by 1,259
  12. 2023Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debatecited by 995
38 more
  1. 2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 965
  2. 2023Teaching Large Language Models to Self-Debugcited by 963
  3. 2020Language (Technology) is Power: A Critical Survey of "Bias" in NLPcited by 951
  4. 2022STaR: Bootstrapping Reasoning With Reasoningcited by 895
  5. 2022Solving math word problems with process- and outcome-based feedbackcited by 871
  6. 2023Reasoning with Language Model is Planning with World Modelcited by 846
  7. 2023Large Language Models as Optimizerscited by 830
  8. 2023SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Modelscited by 787
  9. 2023CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquingcited by 751
  10. 2023Can Large Language Models Be an Alternative to Human Evaluations?cited by 729
  11. 2023ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debatecited by 724
  12. 2022Large Language Models Can Self-Improvecited by 656
  13. 2023Automatic Prompt Optimization with "Gradient Descent" and Beam Searchcited by 628
  14. 2023Active Retrieval Augmented Generationcited by 506
  15. 2023Reinforced Self-Training (ReST) for Language Modelingcited by 486
  16. 2023Logic-LM: Empowering Large Language Models with Symbolic Solvers for Faithful Logical Reasoningcited by 379
  17. 2020Asking and Answering Questions to Evaluate the Factual Consistency of Summariescited by 376
  18. 2022Self-critiquing models for assisting human evaluatorscited by 372
  19. 2023Language Models can Solve Computer Taskscited by 371
  20. 2022CodeT: Code Generation with Generated Testscited by 369
  21. 2023GPTScore: Evaluate as You Desirecited by 357
  22. 2023Check Your Facts and Try Again: Improving Large Language Models with External Knowledge and Automated Feedbackcited by 351
  23. 2022CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learningcited by 342
  24. 2021Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLPcited by 336
  25. 2023Chain-of-Verification Reduces Hallucination in Large Language Modelscited by 317
  26. 2023Analyzing and Mitigating Object Hallucination in Large Vision-Language Modelscited by 314
  27. 2023How Language Model Hallucinations Can Snowballcited by 310
  28. 2022Large Language Models are Better Reasoners with Self-Verificationcited by 304
  29. 2023ReConcile: Round-Table Conference Improves Reasoning via Consensus among Diverse LLMscited by 297
  30. 2020FEQA: A Question Answering Evaluation Framework for Faithfulness Assessment in Abstractive Summarizationcited by 292
  31. 2022RARR: Researching and Revising What Language Models Say, Using Language Modelscited by 286
  32. 2022Generating Sequences by Learning to Self-Correctcited by 271
  33. 2023From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Modelscited by 257
  34. 2023FacTool: Factuality Detection in Generative AI -- A Tool Augmented Framework for Multi-Task and Multi-Domain Scenarioscited by 247
  35. 2021QuestEval: Summarization Asks for Fact-based Evaluationcited by 241
  36. 2023LEVER: Learning to Verify Language-to-Code Generation with Executioncited by 228
  37. 2023LLM Self Defense: By Self Examination, LLMs Know They Are Being Trickedcited by 226
  38. 2023Woodpecker: Hallucination Correction for Multimodal Large Language Modelscited by 223

50 most cited of 113 All 113 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 25; 2025: 85; 2026: 120.

230 papers.

Sort:

By area:

  1. 2024Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMscited by 564
  2. 2024Training Language Models to Self-Correct via Reinforcement Learningcited by 405
  3. 2024ProcessBench: Identifying Process Errors in Mathematical Reasoningcited by 277
  4. 2024Multi-Step Reasoning with Large Language Models, a Surveycited by 140
  5. 2025OverThink: Slowdown Attacks on Reasoning LLMscited by 94
  6. 2025PPTAgent: Generating and Evaluating Presentations Beyond Text-to-Slidescited by 83
  7. 2025Winning Gold at IMO 2025 with a Model-Agnostic Verification-and-Refinement Pipelinecited by 79
  8. 2025Agent-R: Training Language Model Agents to Reflect via Iterative Self-Trainingcited by 74
  9. 2025Revisiting the Test-Time Scaling of o1-like Models: Do they Truly Possess Test-Time Scaling Capabilities?cited by 69
  10. 2025AgentGym-RL: Training LLM Agents for Long-Horizon Decision Making through Multi-Turn Reinforcement Learningcited by 68
  11. 2025Talk Isn't Always Cheap: Understanding Failure Modes in Multi-Agent Debatecited by 63
  12. 2024TICKing All the Boxes: Generated Checklists Improve LLM Evaluation and Generationcited by 62
38 more
  1. 2025The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMscited by 62
  2. 2026Self-Distillation Zero: Self-Revision Turns Binary Rewards into Dense Supervisioncited by 58
  3. 2024Internal Consistency and Self-Feedback in Large Language Models: A Surveycited by 49
  4. 2025Sample, Scrutinize and Scale: Effective Inference-Time Search by Scaling Verificationcited by 49
  5. 2025Large Language Models Hallucination: A Comprehensive Surveycited by 45
  6. 2024Understanding the Dark Side of LLMs' Intrinsic Self-Correctioncited by 42
  7. 2025Correlated Errors in Large Language Modelscited by 42
  8. 2024MAgICoRe: Multi-Agent, Iterative, Coarse-to-Fine Refinement for Reasoningcited by 40
  9. 2024Optima: Optimizing Effectiveness and Efficiency for LLM-Based Multi-Agent Systemcited by 40
  10. 2025A Survey of Test-Time Compute: From Intuitive Inference to Deliberate Reasoningcited by 40
  11. 2025Deconstructing Long Chain-of-Thought: A Structured Reasoning Optimization Framework for Long CoT Distillationcited by 39
  12. 2025S2R: Teaching LLMs to Self-verify and Self-correct via Reinforcement Learningcited by 37
  13. 2024MAGIC: Generating Self-Correction Guideline for In-Context Text-to-SQLcited by 34
  14. 2024LLM Self-Correction with DeCRIM: Decompose, Critique, and Refine for Enhanced Following of Instructions with Multiple Constraintscited by 34
  15. 2024AlphaVerus: Bootstrapping Formally Verified Code Generation through Self-Improving Translation and Treefinementcited by 33
  16. 2025Theoretical Physics Benchmark (TPBench) -- a Dataset and Study of AI Reasoning Capabilities in Theoretical Physicscited by 33
  17. 2024Confidence v.s. Critique: A Decomposition of Self-Correction Capability for LLMscited by 28
  18. 2025LLMs Can Generate a Better Answer by Aggregating Their Own Responsescited by 27
  19. 2025ArcMemo: Abstract Reasoning Composition with Lifelong LLM Memorycited by 22
  20. 2024Toward LLM-Agent-Based Modeling of Transportation Systems: A Conceptual Frameworkcited by 19
  21. 2025Self-Correction Bench: Uncovering and Addressing the Self-Correction Blind Spot in Large Language Modelscited by 19
  22. 2025Zoom-Refine: Boosting High-Resolution Multimodal Understanding via Localized Zoom and Self-Refinementcited by 18
  23. 2025Variation in Verification: Understanding Verification Dynamics in Large Language Modelscited by 18
  24. 2024S^3cMath: Spontaneous Step-level Self-correction Makes Large Language Models Better Mathematical Reasonerscited by 17
  25. 2025An Empirical Study of LLM-as-a-Judge: How Design Choices Impact Evaluation Reliabilitycited by 17
  26. 2025SC-Captioner: Improving Image Captioning with Self-Correction by Reinforcement Learningcited by 16
  27. 2024VASCAR: Content-Aware Layout Generation via Visual-Aware Self-Correctioncited by 14
  28. 2025ProgCo: Program Helps Self-Correction of Large Language Modelscited by 14
  29. 2025Sailing by the Stars: A Survey on Reward Models and Learning Strategies for Learning from Rewardscited by 14
  30. 2025The Energy Cost of Reasoning: Analyzing Energy Usage in LLMs with Test-time Computecited by 14
  31. 2025CulFiT: A Fine-grained Cultural-aware LLM Training Paradigm via Multilingual Critique Data Synthesiscited by 14
  32. 2025Generator-Assistant Stepwise Rollback Framework for Large Language Model Agentcited by 13
  33. 2025Reasoning-as-Logic-Units: Scaling Test-Time Reasoning in Large Language Models Through Logic Unit Alignmentcited by 12
  34. 2025Disambiguate First, Parse Later: Generating Interpretations for Ambiguity Resolution in Semantic Parsingcited by 12
  35. 2025Credible Plan-Driven RAG Method for Multi-Hop Question Answeringcited by 12
  36. 2025Can LLMs Correct Themselves? A Benchmark of Self-Correction in LLMscited by 12
  37. 2025Co-Sight: Enhancing LLM-Based Agents via Conflict-Aware Meta-Verification and Trustworthy Reasoning with Structured Factscited by 12
  38. 2025Feedback Friction: LLMs Struggle to Fully Incorporate External Feedbackcited by 11

50 most cited of 230 All 230 (JSON)