Benchmark Validity

Follow this topic Ask about this topic

Benchmark validity studies whether evaluation tasks measure the intended capability and whether their scores justify claims about real-world performance. Researchers audit task coverage, labels, answer keys, contamination, scoring rules, and agent trajectories, using expert review, controlled interventions, item-level analysis, and psychometric methods such as item response theory. They also test robustness to prompts, data splits, evaluation settings, and benchmark composition, and ask whether rankings predict performance on new tasks or in deployment. A central tension is that standardized benchmarks enable comparison but can reward shortcuts or miss real-world needs, so no single score can establish broad capability without evidence tied to a specific claim.

Latest papers

175 papers20 papers in the last four weeks, down 9% on the four weeks before. 0.1% of all new papers.

Design and Report Benchmarks for Knowledge Work
May 22, 2026Yining Hua, Hongbin Na, Cyrus Ayubcha +1Benchmark Design AI Agent Evaluation
Harvard University · University of Technology Sydney · Stanford University +1
Positional Failures in Long-Context LLMs: A Blind Spot in Reasoning Benchmarks
May 22, 2026Chuyifei Zhang, Hongyu Cui, Xiaowen Huang +1LLM Evaluation Long-Context Language Model Evaluation
Beijing Jiaotong University · Central South University of Forestry and Technology
Decomposing and Measuring Evaluation Awareness
May 21, 2026Changling Li, Terry Jingchen Zhang, Jie Zhang +4LLM Evaluation Language Model Safety Evaluation
ETH Zürich · Max Planck Institute for Intelligent Systems · University of Toronto & Vector Institute +3
Healthcare LLM Benchmarks Are Only as Good as Their Explicit Assumptions
May 21, 2026Naveen Raman, Santiago Cortes-Gomez, Mateo Dulce Rubio +2Healthcare Benchmark Design
New York University
Measuring Security Without Fooling Ourselves: Why Benchmarking Agents Is Hard
May 21, 2026Sahar Abdelnabi, Chris Hicks, Konrad Rieck +1AI Agent Evaluation AI Agent Security
ELLIS Institute Tübingen & MPI-IS & Tübingen AI Center, Germany · The Alan Turing Institute, London, UK · BIFOLD & Technische Universität Berlin, Germany +1
Provable Joint Decontamination for Benchmarking Multiple Large Language Models
May 20, 2026Zhenlong Liu, Hao Zeng, Hongxin WeiLLM Evaluation Benchmark Contamination
Department of Statistics and Data Science, Southern University of Science and Technology · Shanghai Innovation Institute
AgentAtlas: Beyond Outcome Leaderboards for LLM Agents
May 19, 2026Parsa Mazaheri, Kasra MazaheriLLM Agent Reliability Computer-Use Agent Benchmarks
University of California, Santa Cruz · Massachusetts Institute of Technology
LLM Benchmark Datasets Should Be Contamination-Resistant
May 19, 2026Ali Al-Lawati, Jason Lucas, Dongwon Lee +1LLM Evaluation Benchmark Contamination
The Pennsylvania State University, University Park, PA, USA.
Overeager Coding Agents: Measuring Out-of-Scope Actions on Benign Tasks
May 18, 2026Yubin Qu, Ying Zhang, Yanjun Zhang +4Tool Access Control for LLM Agents Computer-Use Agent Benchmarks
Griffith University · Wake Forest University · Nanyang Technological University +2
Are Sparse Autoencoder Benchmarks Reliable?
May 18, 2026David ChaninBenchmark Auditing LLM Interpretability
Decode Research, MATS, UCL
Position: State-of-the-Art Claims Require State-of-the-Art Evidence
May 17, 2026YongKyung OhBenchmark Validity
Medical & Imaging Informatics (MII), University of California, Los Angeles (UCLA)
The Great Pretender: A Stochasticity Problem in LLM Jailbreak
May 14, 2026Jean-Philippe Monteuuis, Cong Chen, Jonathan PetitLLM Safety Benchmarks LLM Evaluation
Qualcomm Technologies, Inc.
The Evaluation Trap: Benchmark Design as Theoretical Commitment
May 13, 2026Theodore J KalaitzidisBenchmark Auditing Benchmark Design
Brown University
Unsteady Metrics and Benchmarking Cultures of AI Model Builders
May 13, 2026Stefan Baack, Christo Buschek, Maty BohacekLLM Evaluation Benchmark Design
Independent Researcher · Mozilla Fellowship · Stanford University
Physics-R1: An Audited Olympiad Corpus and Recipe for Visual Physics Reasoning
May 13, 2026Shan YangVisual Reasoning RL for Language Model Reasoning
Independent Researcher
A Multi-Probe Audit of Clinical-Interview Depression Detection Benchmarks
May 13, 2026Takehiro Ishikawa, Jon DukeDepression Detection Mental Health
College of Computing, Georgia Institute of Technology, Atlanta, GA, USA · Georgia Tech Research Institute, Georgia Institute of Technology, Atlanta, GA, USA
Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack
May 12, 2026Hao Wang, Hanchen Li, Qiuyang Mang +3Computer-Use Agent Benchmarks Reward Hacking
UC Berkeley
The Scaling Law of Evaluation Failure: Why Simple Averaging Collapses Under Data Sparsity and Item Difficulty Gaps, and How Item Response Theory Recovers Ground Truth Across Domains
May 11, 2026Jung Min KangBenchmark Design Benchmark Validity
Independent Researcher · Seoul, South Korea
Can Agent Benchmarks Support Their Scores? Evidence-Supported Bounds for Interactive-Agent Evaluation
May 11, 2026Shanshan Gao, Liyi ZhouComputer-Use Agent Benchmarks Benchmark Design
The University of Sydney
Log analysis is necessary for credible evaluation of AI agents
May 8, 2026Peter Kirgis, Sayash Kapoor, Stephan Rabanser +8Agent Failure Analysis AI Agent Evaluation
Princeton University · Independent · UK AISI +4
Measuring Evaluation-Context Divergence in Open-Weight LLMs: A Paired-Prompt Protocol with Pilot Evidence of Alignment-Pipeline-Specific Heterogeneity
May 7, 2026Florian A. D. Burnat, Brittany I. DavidsonLLM Safety Benchmarks LLM Evaluation
University of Bath
Towards Reliable LLM Evaluation: Correcting the Winner's Curse in Adaptive Benchmarking
May 7, 2026Yang Xu, Jiefu Zhang, Haixiang Sun +3LLM Evaluation Selective Inference
Purdue University · Johns Hopkins University · Purdue University, USA
What Single-Prompt Accuracy Misses: A Multi-Variant Reliability Audit of Language Models
May 3, 2026Ranit Karmakar, Jayita ChatterjeeLLM Evaluation Language Model Calibration
Harvard University, Boston, MA 02115 · Boston, MA 02134
Format Sensitivity Index: Token-Controlled Prompt Wrapper Robustness and Schema Compliance in LLM Benchmarking
May 2, 2026Deep Pankajbhai MehtaPrompt Sensitivity LLM Evaluation
Adobe Inc.
What Makes a Good Terminal-Agent Benchmark Task: A Guideline for Adversarial, Difficult, and Legible Evaluation Design
Apr 30, 2026Ivan BercovichComputer-Use Agent Benchmarks Reward Hacking
BenchGuard: Who Guards the Benchmarks? Automated Auditing of LLM Agent Benchmarks
Apr 27, 2026Xinming Tu, Tianze Wang, Yingzhou +4Benchmark Auditing AI Agent Benchmarks
Minta · Allen School, University of Washington, Seattle, WA, USA · Phylo, Inc., South San Francisco, CA, USA +1
All That Glitters Is Not Audio: Rethinking Text Priors and Audio Reliance in Audio-Language Evaluation
Apr 27, 2026Leonardo Haw-Yang Foo, Chih-Kai Yang, Chen-An Li +2Audio-Language Model Evaluation Modality Reliance in Multimodal Models
National Taiwan University · NTU Artificial Intelligence Center of Research Excellence (NTU AI-CoRE)
RARE: Redundancy-Aware Retrieval Evaluation Framework for High-Similarity Corpora
Apr 21, 2026Hanjun Cho, Jay-Yoon LeeSynthetic Benchmark Generation Benchmark Validity
Allganize · Seoul National University
Edit, But Verify: An Empirical Audit of Instructed Code-Editing Benchmarks
Apr 6, 2026Amir M. Ebrahimi, Gopi Krishnan RajbahadurSoftware Engineering Benchmarks Benchmark Validity
Queen’s University, Kingston, Canada
Are Arabic Benchmarks Reliable? QIMMA's Quality-First Approach to LLM Evaluation
Apr 3, 2026Leen AlQadi, Ahmed Alzubaidi, Mohammed Alyafeai +6LLM Evaluation Benchmark Validity
Technology Innovation Institute, Abu Dhabi, UAE

May 22, 2026J/K move · Enter open · S save

Design and Report Benchmarks for Knowledge Work

Yining Hua, Hongbin Na, Cyrus Ayubcha, Levi Lian

Harvard University · University of Technology Sydney · Stanford University+1

Benchmark Design AI Agent Evaluation Benchmark Validity

Open paper AskPDF arXiv

The development of LLM agents has led to a growing body of work on knowledge-work AI, including coding, research, and healthcare. However, current knowledge-work evaluation and benchmark design still largely follow the logic of traditional NLP tasks. As a result, higher benchmark performance does not reliably show that a system can carry out knowledge work in real-world deployment settings. This paper contributes a three-step approach for making explicit how benchmarked tasks represent the work claims attached to their scores: defining the work activity under evaluation, specifying the tested setting, and scoring the appropriate work product. We review work studies showing that knowledge work is organized through roles and responsibilities, local materials and tools, and artifacts that must remain usable in downstream workflows. We then translate these concerns into benchmark design and reporting guidance, covering how tasks should be mapped to work activities, how tested settings should specify materials, tools, roles, and constraints, and how scoring should focus on the work product left by the system. To name the work activity being evaluated and distinguish it from common benchmark tasks, we derive an inventory of 18 work activities from the O{*}NET occupational task database. We demonstrate the approach through three benchmark case analyses: GDPval, a non-code occupational deliverable benchmark; OfficeQA Pro, a grounded document-analysis benchmark scored by final answers; and APEX-SWE, a software-engineering benchmark with executable scored products. These cases show how benchmark design choices shape the strongest work claim a score can support, and where gaps arise between the benchmarked task, tested setting, scored product, and broader work claim.