# Evaluating Generative AI Systems is a Social Science Measurement Challenge

arXiv: https://arxiv.org/abs/2411.10939 · HTML: https://mlsanity.com/paper/2411.10939/

## Key facts

- Cited by 23 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 17 Nov 2024.

## Built on by (most cited first, 23 of 23)

- [Toward an Evaluation Science for Generative AI Systems](https://mlsanity.com/paper/2503.05336/) (2025)
- [Correlated Errors in Large Language Models](https://mlsanity.com/paper/2506.07962/) (2025)
- [Neither Valid nor Reliable? Investigating the Use of LLMs as Judges](https://mlsanity.com/paper/2508.18076/) (2025)
- [Multi-turn Evaluation of Anthropomorphic Behaviours in Large Language Models](https://mlsanity.com/paper/2502.07077/) (2025)
- [Machine Unlearning Doesn't Do What You Think: Lessons for Generative AI Policy and Research](https://mlsanity.com/paper/2412.06966/) (2024)
- [HumanAgencyBench: Scalable Evaluation of Human Agency Support in AI Assistants](https://mlsanity.com/paper/2509.08494/) (2025)
- [Meaning Is Not A Metric: Using LLMs to make cultural context legible at scale](https://mlsanity.com/paper/2505.23785/) (2025)
- [The Singapore Consensus on Global AI Safety Research Priorities](https://mlsanity.com/paper/2506.20702/) (2025)
- [TALES: A Taxonomy and Analysis of Cultural Representations in LLM-generated Stories](https://mlsanity.com/paper/2511.21322/) (2025)
- [Rigor in AI: Doing Rigorous AI Work Requires a Broader, Responsible AI-Informed Conception of Rigor](https://mlsanity.com/paper/2506.14652/) (2025)
- [Understanding, Protecting, and Augmenting Human Cognition with Generative AI: A Synthesis of the CHI 2025 Tools for Thought Workshop](https://mlsanity.com/paper/2508.21036/) (2025)
- [Social Bias in Popular Question-Answering Benchmarks](https://mlsanity.com/paper/2505.15553/) (2025)
- [Measuring What Matters: Connecting AI Ethics Evaluations to System Attributes, Hazards, and Harms](https://mlsanity.com/paper/2510.10339/) (2025)
- [Distinguishing Task-Specific and General-Purpose AI in Regulation](https://mlsanity.com/paper/2506.17347/) (2025)
- [What is a protest anyway? Codebook conceptualization is still a first-order concern in LLM-era classification](https://mlsanity.com/paper/2510.03541/) (2025)
- [Measurement as Bricolage: Examining How Data Scientists Construct Target Variables for Predictive Modeling Tasks](https://mlsanity.com/paper/2507.02819/) (2025)
- [RLHF May Not Reflect Genuine Preferences](https://mlsanity.com/paper/2604.03238/) (2026)
- [AI-Assisted Systematization for Evaluating GenAI Systems](https://mlsanity.com/paper/2605.26001/) (2026)
- [Measuring Biological Capabilities and Risks of AI Agents](https://mlsanity.com/paper/2606.19899/) (2026)
- [Aligning Recommendations with User Popularity Preferences](https://mlsanity.com/paper/2604.01036/) (2026)
- [From Controlled to the Wild: Evaluation of Pentesting Agents for the Real-World](https://mlsanity.com/paper/2605.10834/) (2026)
- [Healthcare LLM Benchmarks Are Only as Good as Their Explicit Assumptions](https://mlsanity.com/paper/2605.22612/) (2026)
- [LLJ Cards: Best practices for the Use of LLMs as Judges](https://mlsanity.com/paper/2609.24516/) (2026)

## Abstract (arXiv)

Across academia, industry, and government, there is an increasing awareness that the measurement tasks involved in evaluating generative AI (GenAI) systems are especially difficult. We argue that these measurement tasks are highly reminiscent of measurement tasks found throughout the social sciences. With this in mind, we present a framework, grounded in measurement theory from the social sciences, for measuring concepts related to the capabilities, impacts, opportunities, and risks of GenAI systems. The framework distinguishes between four levels: the background concept, the systematized concept, the measurement instrument(s), and the instance-level measurements themselves. This four-level approach differs from the way measurement is typically done in ML, where researchers and practitioners appear to jump straight from background concepts to measurement instruments, with little to no explicit systematization in between. As well as surfacing assumptions, thereby making it easier to understand exactly what the resulting measurements do and do not mean, this framework has two important implications for evaluating evaluations: First, it can enable stakeholders from different worlds to participate in conceptual debates, broadening the expertise involved in evaluating GenAI systems. Second, it brings rigor to operational debates by offering a set of lenses for interrogating the validity of measurement instruments and their resulting measurements.