mlsanityData to 7 Oct 2026

Political Compass or Spinning Arrow? Towards More Meaningful Evaluations for Values and Opinions in Large Language Models

arXiv 2402.16786 ↗First posted 26 Feb 2024cs.CL cs.AIcited by 162

Not on the map. Why?

Abstract (arXiv)

Much recent work seeks to evaluate values and opinions in large language models (LLMs) using multiple-choice surveys and questionnaires. Most of this work is motivated by concerns around real-world LLM applications. For example, politically-biased LLMs may subtly influence society when they are used by millions of people. Such real-world concerns, however, stand in stark contrast to the artificiality of current evaluations: real users do not typically ask LLMs survey questions. Motivated by this discrepancy, we challenge the prevailing constrained evaluation paradigm for values and opinions in LLMs and explore more realistic unconstrained evaluations. As a case study, we focus on the popular Political Compass Test (PCT). In a systematic review, we find that most prior work using the PCT forces models to comply with the PCT's multiple-choice format. We show that models give substantively different answers when not forced; that answers change depending on how models are forced; and that answers lack paraphrase robustness. Then, we demonstrate that models give different answers yet again in a more realistic open-ended answer setting. We distill these findings into recommendations and open challenges in evaluating values and opinions in LLMs.

Key facts. Cited by 162 papers in the map (data through 7 Oct 2026). Builds on 28 of its 45 references that are in the map. First posted on arXiv 26 Feb 2024.

Builds on

28 of 45 references.

References · 28

  1. 2022Training language models to follow instructions with human feedbackcited by 15,712
  2. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  3. 2023Jailbroken: How Does LLM Safety Training Fail?cited by 1,565
  4. 2023Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formattingcited by 778
  5. 2023Whose Opinions Do Language Models Reflect?cited by 759
  6. 2020Aligning AI With Shared Human Valuescited by 706
  7. 2023Large Language Models Are Not Robust Multiple Choice Selectorscited by 454
  8. 2023LMSYS-Chat-1M: A Large-Scale Real-World LLM Conversation Datasetcited by 449
  9. 2023Towards Measuring the Representation of Subjective Global Opinions in Language Modelscited by 424
  10. 2022Using cognitive psychology to understand GPT-3cited by 389
  11. 2021Measuring and Improving Consistency in Pretrained Language Modelscited by 361
  12. 2023Role-Play with Large Language Modelscited by 290
16 more
  1. 2023From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Modelscited by 257
  2. 2021Adversarial GLUE: A Multi-Task Benchmark for Robustness Evaluation of Language Modelscited by 227
  3. 2023Evaluating the Moral Beliefs Encoded in LLMscited by 211
  4. 2023On the Robustness of ChatGPT: An Adversarial and Out-of-distribution Perspectivecited by 208
  5. 2023The political ideology of conversational AI: Converging evidence on ChatGPT's pro-environmental, left-libertarian orientationcited by 183
  6. 2023Embers of Autoregression: Understanding Large Language Models Through the Problem They are Trained to Solvecited by 175
  7. 2022Who is GPT-3? An Exploration of Personality, Values and Demographicscited by 112
  8. 2024The Political Preferences of LLMscited by 83
  9. 2023CValues: Measuring the Values of Chinese Large Language Models from Safety to Responsibilitycited by 80
  10. 2023The Self-Perception and Political Biases of ChatGPTcited by 66
  11. 2024"My Answer is C": First-Token Probabilities Do Not Match Text Answers in Instruction-Tuned Language Modelscited by 66
  12. 2023The Past, Present and Better Future of Feedback Learning in Large Language Models for Subjective Human Preferences and Valuescited by 53
  13. 2023You don't need a personality test to know these models are unreliable: Assessing the Reliability of Large Language Models on Psychometric Instrumentscited by 48
  14. 2023The Shifted and The Overlooked: A Task-oriented Investigation of User-GPT Interactionscited by 45
  15. 2023AI in the Gray: Exploring Moderation Policies in Dialogic Large Language Models vs. Human Answers in Controversial Topicscited by 12
  16. 2023Multilingual Coarse Political Stance Classification of Media. The Editorial Line of a ChatGPT and Bard Newspapercited by 9

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 30; 2025: 70; 2026: 62.

162 papers.

Sort:

By area:

  1. 2024PrivacyLens: Evaluating Privacy Norm Awareness of Language Models in Actioncited by 161
  2. 2024Measuring Political Bias in Large Language Models: What Is Said and How It Is Saidcited by 89
  3. 2024Evaluating Large Language Model Biases in Persona-Steered Generationcited by 78
  4. 2024Large Language Models Reflect the Ideology of their Creatorscited by 56
  5. 2024Auxiliary task demands mask the capabilities of smaller language modelscited by 54
  6. 2025Mind the Value-Action Gap: Do LLMs Act in Alignment with Their Values?cited by 49
  7. 2025Why human-AI relationships need socioaffective alignmentcited by 47
  8. 2025Values in the Wild: Discovering and Analyzing Values in Real-World Language Model Interactionscited by 47
  9. 2024LLM economicus? Mapping the Behavioral Biases of LLMs via Utility Theorycited by 44
  10. 2025The Prompt Makes the Person(a): A Systematic Evaluation of Sociodemographic Persona Prompting for Large Language Modelscited by 40
  11. 2025Multilingual != Multicultural: Evaluating Gaps Between Multilingual Capabilities and Cultural Alignment in LLMscited by 39
  12. 2025The Personality Illusion: Revealing Dissociation Between Self-Reports & Behavior in LLMscited by 35
38 more
  1. 2024Stereotype or Personalization? User Identity Biases Chatbot Recommendationscited by 33
  2. 2025Privacy in Action: Towards Realistic Privacy Mitigation and Evaluation for LLM-Powered Agentscited by 33
  3. 2025Pragmatics in the Era of Large Language Models: A Survey on Datasets, Evaluation, Opportunities and Challengescited by 31
  4. 2025Linear Representations of Political Perspective Emerge in Large Language Modelscited by 30
  5. 2025Break the Checkbox: Challenging Closed-Style Evaluations of Cultural Alignment in LLMscited by 26
  6. 2025Diverse Preference Learning for Capabilities and Alignmentcited by 25
  7. 2024Evaluating Large Language Models with Psychometricscited by 22
  8. 2025IssueBench: Millions of Realistic Prompts for Measuring Issue Bias in LLM Writing Assistancecited by 22
  9. 2024Improving Model Evaluation using SMART Filtering of Benchmark Datasetscited by 20
  10. 2024The Potential and Challenges of Evaluating Attitudes, Opinions, and Values in Large Language Modelscited by 19
  11. 2025The Levers of Political Persuasion with Conversational AIcited by 19
  12. 2024xFinder: Large Language Models as Automated Evaluators for Reliable Evaluationcited by 18
  13. 2025Valid Survey Simulations with Limited Human Data: The Roles of Prompting, Fine-Tuning, and Rectificationcited by 17
  14. 2024Mapping and Influencing the Political Ideology of Large Language Models using Synthetic Personascited by 16
  15. 2024Real or Robotic? Assessing Whether LLMs Accurately Simulate Qualities of Human Responses in Dialoguecited by 15
  16. 2025Only a Little to the Left: A Theory-grounded Measure of Political Bias in Large Language Modelscited by 15
  17. 2024CLEAR: Can Language Models Really Understand Causal Graphs?cited by 14
  18. 2024Algorithmic Fidelity of Large Language Models in Generating Synthetic German Public Opinions: A Case Studycited by 14
  19. 2024Look at the Text: Instruction-Tuned Language Models are More Robust Multiple Choice Selectors than You Thinkcited by 13
  20. 2024Measuring Human and AI Values Based on Generative Psychometrics with Large Language Modelscited by 13
  21. 2024HateDay: Insights from a Global Hate Speech Dataset Representative of a Day on Twittercited by 12
  22. 2024Evidence of a log scaling law for political persuasion with large language modelscited by 11
  23. 2025Measuring Political Preferences in AI Systems: An Integrative Approachcited by 9
  24. 2025Do Psychometric Tests Work for Large Language Models? Evaluation of Tests on Sexism, Racism, and Moralitycited by 9
  25. 2025Value Drifts: Tracing Value Alignment During LLM Post-Trainingcited by 9
  26. 2026One Persona, Many Cues, Different Results: How Sociodemographic Cues Impact LLM Personalizationcited by 9
  27. 2024Are Large Language Models Chameleons? An Attempt to Simulate Social Surveyscited by 8
  28. 2024Bias in the Mirror: Are LLMs opinions robust to their own adversarial attacks ?cited by 8
  29. 2025Benchmarking Large Language Models on Multiple Tasks in Bioinformatics NLP with Promptingcited by 8
  30. 2025How Personality Traits Shape LLM Risk-Taking Behaviourcited by 8
  31. 2025Large Language Models Often Say One Thing and Do Anothercited by 8
  32. 2025Spotting Out-of-Character Behavior: Atomic-Level Evaluation of Persona Fidelity in Open-Ended Generationcited by 8
  33. 2025Generative Exaggeration in LLM Social Agents: Consistency, Bias, and Toxicitycited by 8
  34. 2025Measuring Scalar Constructs in Social Science with LLMscited by 8
  35. 2025Survey Response Generation: Generating Closed-Ended Survey Responses In-Silico with Large Language Modelscited by 8
  36. 2024Cultural Fidelity in Large-Language Models: An Evaluation of Online Language Resources as a Driver of Model Performance in Value Representationcited by 7
  37. 2025The Pluralistic Moral Gap: Understanding Judgment and Value Differences between Humans and Large Language Modelscited by 7
  38. 2025What Is The Political Content in LLMs' Pre- and Post-Training Data?cited by 7

50 most cited of 162 All 162 (JSON)