mlsanityData to 7 Oct 2026

Diminished Diversity-of-Thought in a Standard Large Language Model

arXiv 2302.07267 ↗First posted 13 Feb 2023cs.HC cs.AI cs.CLcited by 43

On the map: LLM reasoning → Reasoning benchmarks and evaluation

Abstract (arXiv)

We test whether Large Language Models (LLMs) can be used to simulate human participants in social-science studies. To do this, we run replications of 14 studies from the Many Labs 2 replication project with OpenAI's text-davinci-003 model, colloquially known as GPT3.5. Based on our pre-registered analyses, we find that among the eight studies we could analyse, our GPT sample replicated 37.5% of the original results and 37.5% of the Many Labs 2 results. However, we were unable to analyse the remaining six studies due to an unexpected phenomenon we call the "correct answer" effect. Different runs of GPT3.5 answered nuanced questions probing political orientation, economic preference, judgement, and moral philosophy with zero or near-zero variation in responses: with the supposedly "correct answer." In one exploratory follow-up study, we found that a "correct answer" was robust to changing the demographic details that precede the prompt. In another, we found that most but not all "correct answers" were robust to changing the order of answer choices. One of our most striking findings occurred in our replication of the Moral Foundations Theory survey results, where we found GPT3.5 identifying as a political conservative in 99.6% of the cases, and as a liberal in 99.3% of the cases in the reverse-order condition. However, both self-reported 'GPT conservatives' and 'GPT liberals' showed right-leaning moral foundations. Our results cast doubts on the validity of using LLMs as a general replacement for human participants in the social sciences. Our results also raise concerns that a hypothetical AI-led future may be subject to a diminished diversity-of-thought.

Key facts. Cited by 43 papers in the map (data through 7 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 9 of its 76 references that are in the map. First posted on arXiv 13 Feb 2023.

Builds on

9 of 76 references.

References · 9

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2019A Survey on Bias and Fairness in Machine Learningcited by 1,911
  3. 2022Out of One, Many: Using Language Models to Simulate Human Samplescited by 737
  4. 2019Release Strategies and the Social Impacts of Language Modelscited by 505
  5. 2022Using cognitive psychology to understand GPT-3cited by 389
  6. 2023The political ideology of conversational AI: Converging evidence on ChatGPT's pro-environmental, left-libertarian orientationcited by 183
  7. 2022Who is GPT-3? An Exploration of Personality, Values and Demographicscited by 112
  8. 2021Algorithmic Amplification of Politics on Twittercited by 38
  9. 2023Differentially Private Natural Language Models: Recent Advances and Future Directionscited by 20

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 4; 2024: 12; 2025: 14; 2026: 13.

43 papers.

Sort:

By area:

  1. 2023Challenges and Applications of Large Language Modelscited by 251
  2. 2025LLM Social Simulations Are a Promising Research Methodcited by 97
  3. 2024The Political Preferences of LLMscited by 83
  4. 2024Wisdom of the Silicon Crowd: LLM Ensemble Prediction Capabilities Rival Human Crowd Accuracycited by 66
  5. 2025Language Model Fine-Tuning on Scaled Survey Data for Predicting Distributions of Public Opinionscited by 64
  6. 2024Benchmarking Distributional Alignment of Large Language Modelscited by 58
  7. 2024Virtual Personas for Language Models via an Anthology of Backstoriescited by 49
  8. 2025SimBench: Benchmarking the Ability of Large Language Models to Simulate Human Behaviorscited by 48
  9. 2025The Homogenizing Effect of Large Language Models on Human Expression and Thoughtcited by 35
  10. 2024Empirical evidence of Large Language Model's influence on human spoken communicationcited by 27
  11. 2024LLM-GLOBE: A Benchmark Evaluating the Cultural Values Embedded in LLM Outputcited by 27
  12. 2026This human study did not involve human subjects: Validating LLM simulations as behavioral evidencecited by 26
31 more
  1. 2025Diverse Preference Learning for Capabilities and Alignmentcited by 25
  2. 2023Large Language Model Prediction Capabilities: Evidence from a Real-World Forecasting Tournamentcited by 24
  3. 2024AI-Augmented Predictions: LLM Assistants Improve Human Forecasting Accuracycited by 21
  4. 2024ValueCompass: A Framework for Measuring Contextual Value Alignment Between Human and LLMscited by 20
  5. 2024Toward LLM-Agent-Based Modeling of Transportation Systems: A Conceptual Frameworkcited by 19
  6. 2024Exploring the Frontiers of LLMs in Psychological Applications: A Comprehensive Reviewcited by 17
  7. 2023Studying and improving reasoning in humans and machinescited by 14
  8. 2025The threat of analytic flexibility in using large language models to simulate human datacited by 13
  9. 2025SaudiCulture: A Benchmark for Evaluating Large Language Models Cultural Competence within Saudi Arabiacited by 8
  10. 2025Aligning LLM with human travel choices: a persona-based embedding learning approachcited by 7
  11. 2025Doubly-Robust LLM-as-a-Judge: Externally Valid Estimation with Imperfect Personascited by 7
  12. 2025Large Language Models as Psychological Simulators: A Methodological Guidecited by 6
  13. 2025Cognitive models can reveal interpretable value trade-offs in language modelscited by 3
  14. 2026AI Agents Alone Are Not (Yet) Sufficient for Social Simulationcited by 3
  15. 2024The creative psychometric item generator: a framework for item generation and validation using large language modelscited by 2
  16. 2026Exploring Diverse Generation Paths via Inference-time Stiefel Activation Steeringcited by 2
  17. 2026Parametric Social Identity Injection and Diversification in Public Opinion Simulationcited by 2
  18. 2026What if AI systems weren't chatbots?cited by 2
  19. 2024The Use of AI-Robotic Systems for Scientific Discoverycited by 1
  20. 2025An AI-Powered Research Assistant in the Lab: A Practical Guide for Text Analysis Through Iterative Collaboration with LLMscited by 1
  21. 2026Expressing Social Emotions: Misalignment Between LLMs and Human Cultural Emotion Normscited by 1
  22. 2023Davinci the Dualist: the mind-body divide in large language models and in human learnerscited by 0
  23. 2025Evidence of conceptual mastery in the application of rules by Large Language Modelscited by 0
  24. 2025AI, Humans, and Data Science: Optimizing Roles Across Workflows and the Workforcecited by 0
  25. 2026From Demographics to Survey Anchors: Evaluating LLM Agents for Modeling Retirement Attitudescited by 0
  26. 2026Will Scaling Improve Social Simulation with LLMs?cited by 0
  27. 2026More Is Not More: What Matters for Diversity in LLM Opinions?cited by 0
  28. 2026Distributional Validity of a Korean Synthetic Persona Panel: Evidence From the Korea Media Panel Surveycited by 0
  29. 2026Item-Mean Surrogates: Why Richer Persona Data Fail to Improve LLMs as Human Surrogatescited by 0
  30. 2026Large Language Models Substantially Compress Well-Being Inequality but Largely Preserve Its Socioeconomic Structurecited by 0
  31. 2026Reach Into The CHOIR: Free-List Elicitation Uncovers Distinct Model Voices in LLM Ensemblescited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026The Limits of Simulated Societies: How Post-Training and Survey Fine-Tuning Erase Cross-Cultural Variance
  2. 2026On the Credibility of Evaluating LLMs using Survey Questions
  3. 2023Reliability Check: An Analysis of GPT-3's Response to Sensitive Topics and Prompt Wording
  4. 2024Confidence in the Reasoning of Large Language Models
  5. 2023Can LLMs Capture Human Preferences?
  6. 2025Emulating Public Opinion: A Proof-of-Concept of AI-Generated Synthetic Survey Responses for the Chilean Case
  7. 2026Measurement Validity in LLM Cultural Alignment
  8. 2025Single LLM Debate, MoLaCE: Mixture of Latent Concept Experts Against Confirmation Bias