mlsanityData to 7 Oct 2026

Position: Evaluating Generative AI Systems Is a Social Science Measurement Challenge

arXiv 2502.00561 ↗First posted 1 Feb 2025cs.CYcited by 61

Not on the map. Why?

Abstract (arXiv)

The measurement tasks involved in evaluating generative AI (GenAI) systems lack sufficient scientific rigor, leading to what has been described as "a tangle of sloppy tests [and] apples-to-oranges comparisons" (Roose, 2024). In this position paper, we argue that the ML community would benefit from learning from and drawing on the social sciences when developing and using measurement instruments for evaluating GenAI systems. Specifically, our position is that evaluating GenAI systems is a social science measurement challenge. We present a four-level framework, grounded in measurement theory from the social sciences, for measuring concepts related to the capabilities, behaviors, and impacts of GenAI systems. This framework has two important implications: First, it can broaden the expertise involved in evaluating GenAI systems by enabling stakeholders with different perspectives to participate in conceptual debates. Second, it brings rigor to both conceptual and operational debates by offering a set of lenses for interrogating validity.

Key facts. Cited by 61 papers in the map (data through 7 Oct 2026). First posted on arXiv 1 Feb 2025.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 12; 2026: 49.

61 papers.

Sort:

By area:

  1. 2025Fairness through Difference Awareness: Measuring Desired Group Discrimination in LLMscited by 20
  2. 2025The Case for "Thick Evaluations" of Cultural Representation in AIcited by 16
  3. 2026Comparison requires valid measurement: Rethinking attack success rate comparisons in AI red teamingcited by 15
  4. 2025The Inadequacy of Offline LLM Evaluations: A Need to Account for Personalization in Model Behaviorcited by 8
  5. 2025Taxonomizing Representational Harms using Speech Act Theorycited by 7
  6. 2025Rigor in AI: Doing Rigorous AI Work Requires a Broader, Responsible AI-Informed Conception of Rigorcited by 7
  7. 2025Social Bias in Popular Question-Answering Benchmarkscited by 5
  8. 2025Sycophantic AI Decreases Prosocial Intentions and Promotes Dependencecited by 5
  9. 2026ProcCtrlBench: Evaluating Process-Level Defects and Control Preservation in LLM Coding Agentscited by 5
  10. 2026Overseeing Agents Without Constant Oversight: Challenges and Opportunitiescited by 3
  11. 2026Position: Coding Benchmarks Are Misaligned with Agentic Software Engineeringcited by 3
  12. 2025We Need to Measure Data Diversity in NLP -- Better and Broadercited by 2
38 more
  1. 2025Reasoning Beyond Labels: Measuring LLM Sentiment in Low-Resource, Culturally Nuanced Contextscited by 2
  2. 2025What is a protest anyway? Codebook conceptualization is still a first-order concern in LLM-era classificationcited by 2
  3. 2025Risk Management for Mitigating Benchmark Failure Modes: BenchRiskcited by 2
  4. 2026Agent Benchmarks Fail Public Sector Requirementscited by 2
  5. 2026From Ground Truth to Measurement: A Statistical Framework for Human Labelingcited by 2
  6. 2026AI as a Tool for Simulation-Based Experiments in Literary Studiescited by 2
  7. 2026CIRCLE: A Framework for Evaluating AI from a Real-World Lenscited by 1
  8. 2026SEED-SET: Scalable Evolving Experimental Design for System-level Ethical Testingcited by 1
  9. 2026Making AI Evaluation Deployment Relevant Through Context Specificationcited by 1
  10. 2026From Use to Oversight: How Mental Models Influence User Behavior and Output in AI Writing Assistantscited by 1
  11. 2026Diversity Without Fidelity: A Solver-Sampler Mismatch in Multi-Agent LLM Negotiation Simulationcited by 1
  12. 2026Bye Bye Perspective API: Lessons for Building and Governing Measurement Infrastructurecited by 1
  13. 2026Offline Preference-Based Trajectory Evaluationcited by 1
  14. 2026What Current AI Benchmarks Leave Unmeasured: Modality, Search, Citations, and Implications (for Safety Evaluations)cited by 1
  15. 2026Do Assessment Instruments Measure the Same Thing for Humans and LLMs? A Latent Structure Analysiscited by 1
  16. 2025Against 'softmaxing' culturecited by 0
  17. 2026Selecting Language Models for Social Science: Start Small, Start Open, and Validatecited by 0
  18. 2026RCTs for Frontier AI Governance: Methodological Challenges and Solutions for Human Uplift Studiescited by 0
  19. 2026Real-World AI Evaluation: How FRAME Generates Systematic Evidence to Resolve the Decision-Maker's Dilemmacited by 0
  20. 2026Questionnaire Responses Do not Capture the Safety of AI Agentscited by 0
  21. 2026BenchBrowser: Retrieving Evidence for Evaluating Benchmark Validitycited by 0
  22. 2026Aligning Recommendations with User Popularity Preferencescited by 0
  23. 2026Towards Apples to Apples for AI Evaluations: From Real-World Use Cases to Evaluation Scenarioscited by 0
  24. 2026Grounding Text Embeddings in Stakeholder Associationscited by 0
  25. 2026Validity Threats for Foundation Model Researchcited by 0
  26. 2026On Defining Erasure Harms for NLPcited by 0
  27. 2026Language Models as Measurement Apparatus for Culturecited by 0
  28. 2026Agent Step Value: Auditing Evaluator-Channel Reversals in Black-Box Agent Tracescited by 0
  29. 2026Validating LLMs in social science: Epistemic threats and emerging normscited by 0
  30. 2026Exposure is not manifestation: measurement target and output resolution jointly determine which behavioural-faithfulness evaluator winscited by 0
  31. 2026AI-accelerated End-to-End Framework for Rapid Professional Upskillingcited by 0
  32. 2026Instrument Effects in Language-Model Honesty Evaluation: An Auditable Single-System Demonstrationcited by 0
  33. 2026Why Public Service AI Governance Frameworks Risk Failing in the Age of General-Purpose AI: Lessons from Policingcited by 0
  34. 2026Long-term Measurements: Towards a Longitudinal Understanding of Human-AI Interactionscited by 0
  35. 2026Emulate or Estimate? The Divergent Strengths of Base and Post-Trained Language Models for Opinion Simulationcited by 0
  36. 2026Causal Behavioral Evaluation of AI Agents at Scale via Automated Behavioral Sciencecited by 0
  37. 2026Position: Reasoning is a Learnable Rule-Based Processcited by 0
  38. 2026ASSERT: A Measurement Pipeline for GenAI Auditscited by 0

50 most cited of 61 All 61 (JSON)