mlsanityData to 7 Oct 2026

Are You Sure? Challenging LLMs Leads to Performance Drops in The FlipFlop Experiment

arXiv 2311.08596 ↗First posted 14 Nov 2023cs.CLcited by 71

Not on the map. Why?

Abstract (arXiv)

The interactive nature of Large Language Models (LLMs) theoretically allows models to refine and improve their answers, yet systematic analysis of the multi-turn behavior of LLMs remains limited. In this paper, we propose the FlipFlop experiment: in the first round of the conversation, an LLM completes a classification task. In a second round, the LLM is challenged with a follow-up phrase like "Are you sure?", offering an opportunity for the model to reflect on its initial answer, and decide whether to confirm or flip its answer. A systematic study of ten LLMs on seven classification tasks reveals that models flip their answers on average 46% of the time and that all models see a deterioration of accuracy between their first and final prediction, with an average drop of 17% (the FlipFlop effect). We conduct finetuning experiments on an open-source LLM and find that finetuning on synthetically created data can mitigate - reducing performance deterioration by 60% - but not resolve sycophantic behavior entirely. The FlipFlop experiment illustrates the universality of sycophantic behavior in LLMs and provides a robust framework to analyze model behavior and evaluate future models.

Key facts. Cited by 71 papers in the map (data through 7 Oct 2026). Builds on 28 of its 48 references that are in the map. First posted on arXiv 14 Nov 2023.

Builds on

28 of 48 references.

References · 28

  1. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  2. 2020Measuring Massive Multitask Language Understandingcited by 7,768
  3. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  4. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,304
  5. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,710
  6. 2017Deep reinforcement learning from human preferencescited by 3,846
  7. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,317
  8. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 2,950
  9. 2021TruthfulQA: Measuring How Models Mimic Human Falsehoodscited by 2,919
  10. 2019BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questionscited by 2,140
  11. 2022Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language modelscited by 1,970
  12. 2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,739
16 more
  1. 2023Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debatecited by 995
  2. 2023Towards Understanding Sycophancy in Language Modelscited by 988
  3. 2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 965
  4. 2022Discovering Language Model Behaviors with Model-Written Evaluationscited by 805
  5. 2023The RefinedWeb Dataset for Falcon LLM: Outperforming Curated Corpora with Web Data, and Web Data Onlycited by 755
  6. 2017Crowdsourcing Multiple Choice Science Questionscited by 742
  7. 2023LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in Large Language Modelscited by 328
  8. 2023REFINER: Reasoning Feedback on Intermediate Representationscited by 203
  9. 2023An Empirical Study of the Non-determinism of ChatGPT in Code Generationcited by 134
  10. 2022Understanding Factual Errors in Summarization: Errors, Summarizers, Datasets, Error Detectorscited by 103
  11. 2022Do Androids Laugh at Electric Sheep? Humor "Understanding" Benchmarks from The New Yorker Caption Contestcited by 98
  12. 2018Complex Sequential Question Answering: Towards Learning to Converse Over Linked Question Answer Pairs with a Knowledge Graphcited by 67
  13. 2023Embrace Divergence for Richer Insights: A Multi-document Summarization Benchmark and a Case Study on Summarizing Diverse Information from News Articlescited by 45
  14. 2023LLMs as Factual Reasoners: Insights from Existing Benchmarks and Beyondcited by 37
  15. 2023On the Intersection of Self-Correction and Trust in Language Modelscited by 12
  16. 2022Discord Questions: A Computational Approach To Diversity Analysis in News Coveragecited by 11

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 1; 2024: 10; 2025: 23; 2026: 37.

71 papers.

Sort:

By area:

  1. 2025LLMs Get Lost In Multi-Turn Conversationcited by 400
  2. 2024To Believe or Not to Believe Your LLMcited by 120
  3. 2024Sycophancy in Large Language Models: Causes and Mitigationscited by 97
  4. 2025Measuring Sycophancy of Language Models in Multi-turn Dialoguescited by 91
  5. 2024How Well Do LLMs Represent Values Across Cultures? Empirical Analysis of LLM Responses Based on Hofstede Cultural Dimensionscited by 51
  6. 2025Flipping the Dialogue: Training and Evaluating User Language Modelscited by 50
  7. 2025A Survey of Test-Time Compute: From Intuitive Inference to Deliberate Reasoningcited by 40
  8. 2025Fostering Appropriate Reliance on Large Language Models: The Role of Explanations, Sources, and Inconsistenciescited by 39
  9. 2025TRUTH DECAY: Quantifying Multi-Turn Sycophancy in Language Modelscited by 31
  10. 2024Prompt Leakage effect and defense strategies for multi-turn LLM interactionscited by 23
  11. 2024PerfCodeGen: Improving Performance of LLM Generated Code with Execution Feedbackcited by 22
  12. 2026Sycophancy Hides Linearly in the Attention Headscited by 21
38 more
  1. 2026How RLHF Amplifies Sycophancycited by 20
  2. 2024Teaching Models to Balance Resisting and Accepting Persuasioncited by 19
  3. 2025Challenging the Evaluator: LLM Sycophancy Under User Rebuttalcited by 18
  4. 2025Sycophancy under Pressure: Evaluating and Mitigating Sycophantic Bias via Adversarial Dialogues in Scientific QAcited by 16
  5. 2024LLM Targeted Underperformance Disproportionately Impacts Vulnerable Userscited by 15
  6. 2024Search Engines in an AI Era: The False Promise of Factual and Verifiable Source-Cited Responsescited by 15
  7. 2025Herd Behavior: Investigating Peer Influence in LLM-based Multi-Agent Systemscited by 15
  8. 2023Self-Contradictory Reasoning Evaluation and Detectioncited by 14
  9. 2024Understanding the Effects of Iterative Prompting on Truthfulnesscited by 14
  10. 2025Calibrating Verbalized Confidence with Self-Generated Distractorscited by 14
  11. 2026What Counts as AI Sycophancy? A Taxonomy and Expert Survey of a Fragmented Constructcited by 12
  12. 2025BASIL: Bayesian Assessment of Sycophancy in LLMscited by 11
  13. 2025DeepTRACE: Auditing Deep Research AI Systems for Tracking Reliability Across Citations and Evidencecited by 10
  14. 2025MONICA: Real-Time Monitoring and Calibration of Chain-of-Thought Sycophancy in Large Reasoning Modelscited by 10
  15. 2026The Fragility Of Moral Judgment In Large Language Modelscited by 8
  16. 2024Are LLMs classical or nonmonotonic reasoners? Lessons from genericscited by 7
  17. 2026Good Arguments Against the People Pleasers: How Reasoning Mitigates (Yet Masks) LLM Sycophancycited by 7
  18. 2025B-score: Detecting biases in large language models using response historycited by 6
  19. 2025Time-To-Inconsistency: A Survival Analysis of Large Language Model Robustness to Adversarial Attackscited by 6
  20. 2025Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewardscited by 6
  21. 2026LLM Spirals of Delusion: A Benchmarking Audit Study of AI Chatbot Interfacescited by 6
  22. 2026SycoEval-EM: Sycophancy Evaluation of Large Language Models in Simulated Clinical Encounters for Emergency Carecited by 5
  23. 2026Too Polite to Disagree: Understanding Sycophancy Propagation in Multi-Agent Systemscited by 5
  24. 2025PENDULUM: A Benchmark for Assessing Sycophancy in Multimodal Large Language Modelscited by 4
  25. 2025Eliciting Behaviors in Multi-Turn Conversationscited by 4
  26. 2026Consistency of Large Reasoning Models Under Multi-Turn Attackscited by 4
  27. 2026Personalization Increases Affective Alignment but Has Role-Dependent Effects on Epistemic Independence in LLMscited by 4
  28. 2026Most LLM Conformity Needs No Speaker: Measuring the Speaker-Free Floor in Peer-Pressure Benchmarkscited by 4
  29. 2025LLM-driven Constrained Copy Generation through Iterative Refinementcited by 3
  30. 2025All Roads Lead to Rome: Graph-Based Confidence Estimation for Large Language Model Reasoningcited by 3
  31. 2025Interpreting and Mitigating Unwanted Uncertainty in LLMscited by 2
  32. 2026Overalignment in Frontier LLMs: An Empirical Study of Sycophantic Behaviour in Healthcarecited by 2
  33. 2026Certainty robustness: Evaluating LLM stability under self-challenging promptscited by 2
  34. 2026Reasoning Gets Harder for LLMs Inside A Dialoguecited by 2
  35. 2026How LLMs Are Persuaded: A Few Attention Heads, Reroutedcited by 2
  36. 2025AI Through the Human Lens: Investigating Cognitive Theories in Machine Psychologycited by 1
  37. 2026Are LLMs Vulnerable to Preference-Undermining Attacks (PUA)? A Factorial Analysis Methodology for Diagnosing the Trade-off between Preference Alignment and Real-World Validitycited by 1
  38. 2026Conformal Prediction for Risk-Controlled Medical Entity Extraction Across Clinical Domainscited by 1

50 most cited of 71 All 71 (JSON)