mlsanityData to 7 Oct 2026

CoMPosT: Characterizing and Evaluating Caricature in LLM Simulations

arXiv 2310.11501 ↗First posted 17 Oct 2023cs.CL cs.AI cs.CYcited by 94

Not on the map. Why?

Abstract (arXiv)

Recent work has aimed to capture nuances of human behavior by using LLMs to simulate responses from particular demographics in settings like social science experiments and public opinion surveys. However, there are currently no established ways to discuss or evaluate the quality of such LLM simulations. Moreover, there is growing concern that these LLM simulations are flattened caricatures of the personas that they aim to simulate, failing to capture the multidimensionality of people and perpetuating stereotypes. To bridge these gaps, we present CoMPosT, a framework to characterize LLM simulations using four dimensions: Context, Model, Persona, and Topic. We use this framework to measure open-ended LLM simulations' susceptibility to caricature, defined via two criteria: individuation and exaggeration. We evaluate the level of caricature in scenarios from existing work on LLM simulations. We find that for GPT-4, simulations of certain demographics (political and marginalized groups) and topics (general, uncontroversial) are highly susceptible to caricature.

Key facts. Cited by 94 papers in the map (data through 7 Oct 2026). Builds on 34 of its 88 references that are in the map. First posted on arXiv 17 Oct 2023.

Builds on

34 of 88 references.

References · 34

  1. 2022Training language models to follow instructions with human feedbackcited by 15,712
  2. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,049
  3. 2023Generative Agents: Interactive Simulacra of Human Behaviorcited by 3,500
  4. 2018Datasheets for Datasetscited by 1,371
  5. 2018Model Cards for Model Reportingcited by 1,068
  6. 2023Whose Opinions Do Language Models Reflect?cited by 759
  7. 2023AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedbackcited by 738
  8. 2022Out of One, Many: Using Language Models to Simulate Human Samplescited by 737
  9. 2022Using Large Language Models to Simulate Multiple Humans and Replicate Human Subject Studiescited by 506
  10. 2022Using cognitive psychology to understand GPT-3cited by 389
  11. 2023The Curse of Recursion: Training on Generated Data Makes Models Forgetcited by 343
  12. 2022Social Simulacra: Creating Populated Prototypes for Social Computing Systemscited by 300
22 more
  1. 2023Marked Personas: Using Natural Language Prompts to Measure Stereotypes in Language Modelscited by 204
  2. 2018WikiHow: A Large Scale Text Summarization Datasetcited by 203
  3. 2022Fine-tuning language models to find agreement among humans with diverse preferencescited by 189
  4. 2021Process for Adapting Language Models to Society (PALMS) with Values-Targeted Datasetscited by 179
  5. 2020Question and Answer Test-Train Overlap in Open-Domain Question Answering Datasetscited by 137
  6. 2019Towards a Critical Race Methodology in Algorithmic Fairnesscited by 124
  7. 2023Artificial Artificial Artificial Intelligence: Crowd Workers Widely Use Large Language Models for Text Production Taskscited by 102
  8. 2023NLPositionality: Characterizing Design Biases of Datasets and Modelscited by 91
  9. 2023To Repeat or Not To Repeat: Insights from Scaling LLM under Token-Crisiscited by 88
  10. 2021Memorization vs. Generalization: Quantifying Data Leakage in NLP Performance Evaluationcited by 69
  11. 2022Towards Intersectionality in Machine Learning: Including More Identities, Handling Underrepresentation, and Performing Evaluationcited by 59
  12. 2023Blind Judgement: Agent-Based Supreme Court Modelling With GPTcited by 37
  13. 2021Do Humans Trust Advice More if it Comes from AI? An Analysis of Human-AI Interactionscited by 35
  14. 2022Theory-Grounded Measurement of U.S. Social Stereotypes in English Language Modelscited by 35
  15. 2017Measuring the Similarity of Sentential Arguments in Dialogcited by 23
  16. 2022Investigations of Performance and Bias in Human-AI Teamwork in Hiringcited by 16
  17. 2020Deceptive AI Explanations: Creation and Detectioncited by 14
  18. 2022Discovering Differences in the Representation of People using Contextualized Semantic Axescited by 12
  19. 2023ACROCPoLis: A Descriptive Framework for Making Sense of Fairnesscited by 8
  20. 2020Stance Prediction for Contemporary Issues: Data and Experimentscited by 7
  21. 2022Prescriptive and Descriptive Approaches to Machine-Learning Transparencycited by 7
  22. 2021Social Norm Bias: Residual Harms of Fairness-Aware Algorithmscited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 5; 2024: 33; 2025: 34; 2026: 22.

94 papers.

Sort:

By area:

  1. 2023SOTOPIA: Interactive Evaluation for Social Intelligence in Language Agentscited by 350
  2. 2023Large Language Models Empowered Agent-based Modeling and Simulation: A Survey and Perspectivescited by 253
  3. 2024Quantifying the Persona Effect in LLM Simulationscited by 164
  4. 2023Simulating Opinion Dynamics with Networks of LLM-based Agentscited by 153
  5. 2024Best Practices and Lessons Learned on Synthetic Datacited by 104
  6. 2024Systematic Biases in LLM Simulations of Debatescited by 98
  7. 2025LLM Social Simulations Are a Promising Research Methodcited by 97
  8. 2024Evaluating Large Language Model Biases in Persona-Steered Generationcited by 78
  9. 2024Roleplay-doh: Enabling Domain-Experts to Create LLM-simulated Patients via Eliciting and Adhering to Principlescited by 72
  10. 2025Language Model Fine-Tuning on Scaled Survey Data for Predicting Distributions of Public Opinionscited by 64
  11. 2023The Challenge of Using LLMs to Simulate Human Behavior: A Causal Inference Perspectivecited by 57
  12. 2024Virtual Personas for Language Models via an Anthology of Backstoriescited by 49
38 more
  1. 2024BIG5-CHAT: Shaping LLM Personalities Through Training on Human-Grounded Datacited by 49
  2. 2024Random Silicon Sampling: Simulating Human Sub-Population Opinion Using a Large Language Model Based on Group-Level Demographic Informationcited by 48
  3. 2025SimBench: Benchmarking the Ability of Large Language Models to Simulate Human Behaviorscited by 48
  4. 2025The Prompt Makes the Person(a): A Systematic Evaluation of Sociodemographic Persona Prompting for Large Language Modelscited by 40
  5. 2024Benchmarks as Microscopes: A Call for Model Metrologycited by 33
  6. 2024Large Language Models Portray Socially Subordinate Groups as More Homogeneous, Consistent with a Bias Observed in Humanscited by 31
  7. 2024LLMs generate structurally realistic social networks but overestimate political homophilycited by 31
  8. 2025Beyond Demographics: Fine-tuning Large Language Models to Predict Individuals' Subjective Text Perceptionscited by 31
  9. 2025ChatBench: From Static Benchmarks to Human-AI Evaluationcited by 31
  10. 2024AnthroScore: A Computational Linguistic Measure of Anthropomorphismcited by 27
  11. 2026This human study did not involve human subjects: Validating LLM simulations as behavioral evidencecited by 26
  12. 2024Modeling Human Subjectivity in LLMs Using Explicit and Implicit Human Factors in Personascited by 23
  13. 2024Annotation alignment: Comparing LLM and human annotations of conversational safetycited by 22
  14. 2024Helpful assistant or fruitful facilitator? Investigating how personas affect language model behaviorcited by 21
  15. 2025Population-Aligned Persona Generation for LLM-based Social Simulationcited by 20
  16. 2024The Potential and Challenges of Evaluating Attitudes, Opinions, and Values in Large Language Modelscited by 19
  17. 2024Very Large-Scale Multi-Agent Simulation in AgentScopecited by 18
  18. 2025Cultural Learning-Based Culture Adaptation of Language Modelscited by 16
  19. 2024LLM Targeted Underperformance Disproportionately Impacts Vulnerable Userscited by 15
  20. 2024GermanPartiesQA: Benchmarking Commercial Large Language Models and AI Companions for Political Alignment and Sycophancycited by 14
  21. 2024Social Skill Training with Large Language Modelscited by 13
  22. 2023How Far Are LLMs from Believable AI? A Benchmark for Evaluating the Believability of Human Behavior Simulationcited by 11
  23. 2024Exploring Changes in Nation Perception with Nationality-Assigned Personas in LLMscited by 11
  24. 2025LLM-Based Social Simulations Require a Boundarycited by 11
  25. 2024Social Intelligence Data Infrastructure: Structuring the Present and Navigating the Futurecited by 10
  26. 2025SOTOPIA-S4: a user-friendly system for flexible, customizable, and large-scale social simulationcited by 10
  27. 2024Persona Setting Pitfall: Persistent Outgroup Biases in Large Language Models Arising from Social Identity Adoptioncited by 9
  28. 2024Towards Foundation-model-based Multiagent System to Accelerate AI for Social Impactcited by 9
  29. 2025SPHERE: An Evaluation Card for Human-AI Systemscited by 9
  30. 2025Improving Language Model Personas via Rationalization with Psychological Scaffoldscited by 9
  31. 2024Mitigating Propensity Bias of Large Language Models for Recommender Systemscited by 8
  32. 2024ELLMA-T: an Embodied LLM-agent for Supporting English Language Learning in Social VRcited by 8
  33. 2025MultiHoax: A Dataset of Multi-hop False-Premise Questionscited by 8
  34. 2025Generative Exaggeration in LLM Social Agents: Consistency, Bias, and Toxicitycited by 8
  35. 2025Systematic Characterization of LLM Quantization: A Performance, Energy, and Quality Perspectivecited by 8
  36. 2024Demystifying the Communication Characteristics for Distributed Transformer Modelscited by 7
  37. 2026Assessing the Reliability of Persona-Conditioned LLMs as Synthetic Survey Respondentscited by 7
  38. 2024Vision-Language Models Generate More Homogeneous Stories for Phenotypically Black Individualscited by 6

50 most cited of 94 All 94 (JSON)