PersonaCUABench: Benchmarking Computer-Use Agents Serving Diverse Personas

Haoming Li; Defu Cao; Xu Hu; Yingtao Luo; Yiyang Wang; Xianghe Xu; Jiashu He; Wanting Jiang; Yifan Zhang; Yunhan Zhao; Siyang Wu; Chen Yang; Haixin Wang; Zheyuan Deng; Taoran Li; Jingwen Zhang; Zhejian Peng; Shushu Liang; Yue Zhao; Bingzhe Li; Feng Chen; Yifan Wang; Jintao Huang; Yijun Wang; Yuexing Hao; Xiaomin Li

arXiv · 2610.22990

PersonaCUABench: Benchmarking Computer-Use Agents Serving Diverse Personas

View original source ↗Read source document ↗Export metadata ↓

Abstract

Computer-use agents (CUAs) are commonly evaluated against fixed task specifications, but users may omit requirements, add constraints, or require approval before consequential actions. Completing the benchmark instruction does not guarantee serving the person who asked. We introduce PersonaCUABench, which holds 74 tasks across 11 domains fixed and varies only who asks. A persona agent, playing one of 44 personas in four groups, rewrites each instruction in that user's own words and answers the CUA's questions, while the CUA sees only the dialogue, never the profile or the original instruction. Across nine backbones (3,330 runs), mean task completion falls by 36% relative to the original requests. The losses have two sources: in 59.9% of cells that lose points, the agent engages with what the user asked for and still gets it wrong, and 11.2% are stops the user requested that the checker scores as failures. Even when a persona request is fully completed, the user gives the top satisfaction rating in only 62.7% of cells, against 84.1% for the original requests. CUAs should therefore be evaluated not only on task completion but also on whether they meet the requirements and constraints users express. Code: https://github.com/MatrAIx-ai/PersonaCUABench. Project page: https://matraix.ai/research/persona-cua.html

Explore related subjects

cs.HC cs.AI cs.CY cs.LG

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Haoming Li, Defu Cao, Xu Hu, Yingtao Luo, Yiyang Wang, Xianghe Xu, Jiashu He, Wanting Jiang, Yifan Zhang, Yunhan Zhao, Siyang Wu, Chen Yang, Haixin Wang, Zheyuan Deng, Taoran Li, Jingwen Zhang, Zhejian Peng, Shushu Liang, Yue Zhao, Bingzhe Li, Feng Chen, Yifan Wang, Jintao Huang, Yijun Wang, Yuexing Hao, Xiaomin Li. 2026-10-16. PersonaCUABench: Benchmarking Computer-Use Agents Serving Diverse Personas. https://arxiv.org/abs/2610.22990

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Designing Human-AI Collaborative Science: The 9A Framework

Artificial intelligence (AI) for Science is often described as a progression from human-directed assistance toward increasingly autonomous AI scientists. Human-AI collaboration offers a further mode in which human and AI contributions substantively shape scientific reasoning, judgments, and the course of inquiry. Existing research addresses work allocation, human-AI coordination, and system evaluation, but how these aspects interact to shape collaborative scientific systems remains insufficiently articulated. We develop the 9A Framework for Human-AI Collaborative Science (HACS) to connect these considerations across three layers. The organization layer specifies what scientific activities are required (Activity), who performs them (Actor), and who holds authority over consequential decisions (Authority). The coordination layer addresses how participants access relevant research information (Awareness), how scientific work advances through iterative proposal, execution, examination, revision, and resolution (Accountability), and when collaborative arrangements should be adjusted (Adaptability). The evaluation layer establishes what constitutes effective system performance (Achievement), how comparisons can be designed to identify the effects of organizational and coordination choices (Attribution), and what evidence supports the resulting conclusions (Assurance). Together, these layers frame HACS as a system-design problem, providing a basis for designing collaborative scientific workflows and for empirically comparing their effectiveness.

cs.HC↗

arXiv2026

The Ethics of Prompt Engineering

Prompt engineering is commonly understood as a technical practice aimed at eliciting useful, accurate, and appropriately structured outputs from large language models (LLMs). Yet prompts do more than optimize model performance: they shape what information is elicited, which assumptions are introduced, and which perspectives are privileged. In this paper, we argue that prompting should therefore be understood as an epistemic and normative act. We distinguish between two conceptions of prompting: knowledge generation, in which the model is implicitly treated as an epistemic authority, and knowledge injection, in which users provide relevant knowledge, evidence, and context that the model can organize, articulate, and extend. We contend that the latter provides a more appropriate model for responsible LLM use, since prompting cannot substitute for subject-matter expertise or the critical evaluation of model outputs. At the same time, prompts inevitably inject values, intentions, and assumptions into the generation process. Building on these premises, we propose five principles for responsible prompting: epistemic humility, explicit engagement with uncertainty and epistemic underdetermination, source sensitivity, avoidance of deceptive framing, and recognition of model limitations and appropriate task allocation. We further argue that responsibility for AI-generated outputs is distributed across users, developers, institutions, and societies, and that some applications cannot be rendered ethical through better prompting alone.

cs.HC↗

arXiv2026

Immersive Focus+Context Techniques to Assist in Interpretation of High Density Forest Point Clouds

Virtual Reality (VR) is changing how we interact with complex spatial datasets. VR has the potential within forestry to help analyse remotely sensed spatial data for inventory management. However, challenges of scale and legibility of dense forest point clouds must be addressed. To do so, we present three VR Focus+Context (F+C) techniques: Wedgelight, Remote Lantern, and Ray Lantern. These techniques present point cloud data using a constrained highdetail focus within a broader abstracted context representation, manipulated using egocentric and exocentric interactions. We conducted two user studies. The first informed the design of the F+C techniques. The results from the first study indicate that VR significantly outperforms desktop displays in basic analysis of forestry point clouds, with improvements to both time and accuracy. The second study evaluated the performance of the resulting F+C techniques in a task analogous to real-world forestry operations. Our F+C techniques significantly outperformed current approaches, confirming their viability for facilitating detailed and effective forest assessments. These findings indicate that VR has significant advantages for spatial data analysis tasks in forestry, making a strong argument for its adoption in future analytical frameworks. The F+C visualisation techniques we introduced show considerable promise for analysis of high density point clouds, opening new avenues for development in immersive analytics. Supplemental materials (code, scans, results) are available at https://osf.io/hqc9u.

cs.HC↗