# ChatBench: From Static Benchmarks to Human-AI Evaluation

arXiv: https://arxiv.org/abs/2504.07114 · HTML: https://mlsanity.com/paper/2504.07114/

## Key facts

- Cited by 34 papers in the map (data through 14 Oct 2026).
- Builds on 28 of its 52 references that are in the map.
- First posted on arXiv 22 Mar 2025.

## Builds on (most cited first, 28 of 28)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering](https://mlsanity.com/paper/1809.09600/) (2018)
- [Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference](https://mlsanity.com/paper/2403.04132/) (2024)
- [Out of One, Many: Using Language Models to Simulate Human Samples](https://mlsanity.com/paper/2209.06899/) (2022)
- [AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedback](https://mlsanity.com/paper/2305.14387/) (2023)
- [WildChat: 1M ChatGPT Interaction Logs in the Wild](https://mlsanity.com/paper/2405.01470/) (2024)
- [LMSYS-Chat-1M: A Large-Scale Real-World LLM Conversation Dataset](https://mlsanity.com/paper/2309.11998/) (2023)
- [Does the Whole Exceed its Parts? The Effect of AI Explanations on Complementary Team Performance](https://mlsanity.com/paper/2006.14779/) (2020)
- [WildBench: Benchmarking LLMs with Challenging Tasks from Real Users in the Wild](https://mlsanity.com/paper/2406.04770/) (2024)
- [Marked Personas: Using Natural Language Prompts to Measure Stereotypes in Language Models](https://mlsanity.com/paper/2305.18189/) (2023)
- [MediQ: Question-Asking LLMs and a Benchmark for Reliable Interactive Clinical Reasoning](https://mlsanity.com/paper/2406.00922/) (2024)
- [Adding Error Bars to Evals: A Statistical Approach to Language Model Evaluations](https://mlsanity.com/paper/2411.00640/) (2024)
- [Large language models that replace human participants can harmfully misportray and flatten identity groups](https://mlsanity.com/paper/2402.01908/) (2024)
- [Aligning Language Models to User Opinions](https://mlsanity.com/paper/2305.14929/) (2023)
- [CoMPosT: Characterizing and Evaluating Caricature in LLM Simulations](https://mlsanity.com/paper/2310.11501/) (2023)
- [Evaluating Language Models for Mathematics through Interactions](https://mlsanity.com/paper/2306.01694/) (2023)
- [Language Model Fine-Tuning on Scaled Survey Data for Predicting Distributions of Public Opinions](https://mlsanity.com/paper/2502.16761/) (2025)
- [Collaborative Gym: A Framework for Enabling and Evaluating Human-Agent Collaboration](https://mlsanity.com/paper/2412.15701/) (2024)
- [Towards Responsible Development of Generative AI for Education: An Evaluation-Driven Approach](https://mlsanity.com/paper/2407.12687/) (2024)
- [Shaping Human-AI Collaboration: Varied Scaffolding Levels in Co-writing with Language Models](https://mlsanity.com/paper/2402.11723/) (2024)
- [PlatoLM: Teaching LLMs in Multi-Round Dialogue via a User Simulator](https://mlsanity.com/paper/2308.11534/) (2023)
- [BASES: Large-scale Web Search User Simulation with Large Language Model based Agents](https://mlsanity.com/paper/2402.17505/) (2024)
- [IQA-EVAL: Automatic Evaluation of Human-Model Interactive Question Answering](https://mlsanity.com/paper/2408.13545/) (2024)
- [More than Marketing? On the Information Value of AI Benchmarks for Practitioners](https://mlsanity.com/paper/2412.05520/) (2024)
- [Characterizing the Global Crowd Workforce: A Cross-Country Comparison of Crowdworker Demographics](https://mlsanity.com/paper/1812.05948/) (2018)

## Built on by (most cited first, 30 of 34)

- [LLMs Get Lost In Multi-Turn Conversation](https://mlsanity.com/paper/2505.06120/) (2025)
- [Lost in Simulation: LLM-Simulated Users are Unreliable Proxies for Human Users in Agentic Evaluations](https://mlsanity.com/paper/2601.17087/) (2026)
- [HumanLM: Simulating Users with State Alignment Beats Response Imitation](https://mlsanity.com/paper/2603.03303/) (2026)
- [User Feedback in Human-LLM Dialogues: A Lens to Understand Users But Noisy as a Learning Signal](https://mlsanity.com/paper/2507.23158/) (2025)
- [HorizonBench: Long-Horizon Personalization with Evolving Preferences](https://mlsanity.com/paper/2604.17283/) (2026)
- [Potemkin Understanding in Large Language Models](https://mlsanity.com/paper/2506.21521/) (2025)
- [Generative Value Conflicts Reveal LLM Priorities](https://mlsanity.com/paper/2509.25369/) (2025)
- [Quantifying the Utility of User Simulators for Building Collaborative LLM Assistants](https://mlsanity.com/paper/2605.09808/) (2026)
- [LLMs Outperform Experts on Challenging Biology Benchmarks](https://mlsanity.com/paper/2505.06108/) (2025)
- [The Inadequacy of Offline LLM Evaluations: A Need to Account for Personalization in Model Behavior](https://mlsanity.com/paper/2509.19364/) (2025)
- [MatrAIx: Simulating the World with 8.3 Billion Persona Agents](https://mlsanity.com/paper/2608.04205/) (2026)
- [Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewards](https://mlsanity.com/paper/2510.18731/) (2025)
- [DiscoverLLM: From Executing Intents to Discovering Them](https://mlsanity.com/paper/2602.03429/) (2026)
- [Exploiting Leaderboards for Large-Scale Distribution of Malicious Models](https://mlsanity.com/paper/2507.08983/) (2025)
- [ELI-Why: Evaluating the Pedagogical Utility of Language Model Explanations](https://mlsanity.com/paper/2506.14200/) (2025)
- [Understanding Persuasive Interactions between Generative Social Agents and Humans: The Knowledge-based Persuasion Model (KPM)](https://mlsanity.com/paper/2602.11483/) (2026)
- [GAUGE: When Not to Trust LLM-as-a-Judge in User-Simulated Evaluation of Task-Oriented Agents](https://mlsanity.com/paper/2609.12191/) (2026)
- [Cultural Compass: A Framework for Organizing Societal Norms to Detect Violations in Human-AI Conversations](https://mlsanity.com/paper/2601.07973/) (2026)
- [Synthetic Users, Real Differences: an Evaluation Framework for User Simulation in Multi-Turn Conversations](https://mlsanity.com/paper/2605.02624/) (2026)
- [ReIn: Conversational Error Recovery with Reasoning Inception](https://mlsanity.com/paper/2602.17022/) (2026)
- [Does Theory of Mind Improvement Really Benefit Human-AI Interactions? Empirical Findings from Interactive Evaluations](https://mlsanity.com/paper/2605.15205/) (2026)
- [Enjoy Your Talk: A Human-Centered Benchmark for Multi-Turn Dialogue with Decoupled User Simulation, Target Modeling, and Judging](https://mlsanity.com/paper/2607.10428/) (2026)
- [KnowSim: Evaluating Information Calibration in LLM Assistants with User Simulators that Learn](https://mlsanity.com/paper/2608.17150/) (2026)
- [CentaurBench: Benchmarking LLM Capabilities on Augmenting vs. Automating Real-World Work Tasks](https://mlsanity.com/paper/2608.18554/) (2026)
- [From Task Success to Productive Success: Evaluating Human-AI Collaboration by Quality and Cost](https://mlsanity.com/paper/2609.21117/) (2026)
- [A Vision for Multisensory Intelligence: Sensing, Science, and Synergy](https://mlsanity.com/paper/2601.04563/) (2026)
- [From Feelings to Metrics: Understanding and Formalizing How Users Vibe-Test LLMs](https://mlsanity.com/paper/2604.14137/) (2026)
- [API Benchmark Scores Do Not Reliably Transfer to Chatbot Interfaces](https://mlsanity.com/paper/2609.08861/) (2026)
- [ORQA: An Occupation-Realistic Question and Answer Framework for LLM Professional Knowledge](https://mlsanity.com/paper/2609.12366/) (2026)
- [LLMs as Oracles: Reliance on LLMs for Subjective Personal Questions](https://mlsanity.com/paper/2609.14849/) (2026)

## Abstract (arXiv)

With the rapid adoption of LLM-based chatbots, there is a pressing need to evaluate what humans and LLMs can achieve together. However, standard benchmarks, such as MMLU, measure LLM capabilities in isolation (i.e., "AI-alone"). Here, we design and conduct a user study to convert MMLU questions into user-AI conversations, by seeding the user with the question and having them carry out a conversation with the LLM to answer their question. We release ChatBench, a new dataset with AI-alone, user-alone, and user-AI data for 396 questions and two LLMs, including 144K answers and 7,336 user-AI conversations. We find that AI-alone accuracy fails to predict user-AI accuracy, with significant differences across multiple subjects (math, physics, and moral reasoning), and we analyze the user-AI conversations to provide insight into how they diverge from AI-alone benchmarks. Finally, we show that fine-tuning a user simulator on a subset of ChatBench improves its ability to estimate user-AI accuracies, increasing correlation on held-out questions by more than 20 points, creating possibilities for scaling interactive evaluation.