[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-83348-en":3,"doc-seo-83348-105":29,"detail-sidebar-cat-0-en-105":83},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},83348,687197207919,"Theodora","https://ap-avatar.wpscdn.com/avatar/a000253d6f5f7c60be?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779446848396160552",8,"Research & Report","Playing ZendoWorld: Challenging AI Agents on Active Visual Concept Induction","ZendoWorld addresses the core challenge of enabling AI agents to jointly perceive complex visual inputs, induce hypotheses about hidden logical rules, and design informative experiments via active scene generation. The work proposes a controlled multi-turn environment inspired by the inductive logic game Zendo, combining raw visual observations with a DSL and game-feedback. Experiments across VLM reasoning, Bayesian particle filtering, dynamic concept discovery, and neuro-symbolic approaches show that high label prediction accuracy does not guarantee rule recovery, revealing distinct perception and induction bottlenecks and weak experimental informativeness.","arXiv :2607 .08233v 1 [ cs .AI] 9 Jul 2026  \nPlaying ZendoWorld: Challenging AI Agents on Active Visual Concept Induction  \nSophia Koehler 1 ,2 Antonia Wüst 1 Inga Ibs3 ,4 Wasu Top Piriyakulkij5 Wolfgang Stammer6 Constantin Rothkopf2 ,3 ,4 Kevin Ellis5  \nKristian Kersting 1 ,2 ,4 ,7  \n1AIML Lab, TU Darmstadt; 2Hessian Center for AI (hessian.AI);  \n3 Psychology of Information Processing, TU Darmstadt; 4 Centre for Cognitive Science, TU Darmstadt; 5 Cornell University;6Max Planck Institute for Informatics, SIC; 7 German Center for AI (DFKI);  \nAbstract  \nA central challenge in building intelligent systems is enabling agents to jointly  \nperceive complex inputs, form hypotheses about hidden patterns, and design infor  \nmative experiments to test them. To study this problem, we propose ZendoWorld,  \na controlled interactive environment in which agents must infer a logical rule about  \nvisual game observations, acquire information by proposing new scenes, and refine  \ntheir hypotheses based on feedback from the game environment. We evaluate  \nseveral agents spanning pure VLM reasoning, Bayesian particle filtering, dynamic  \nconcept discovery, and neuro-symbolic methods. Our main findings are: (1) high  \naccuracy in predicting labels for observed examples does not imply recovery of the  \nunderlying rule; (2) perception and induction are distinct bottlenecks for different  \nagent classes; and (3) VLM-based agents propose near-uninformative experiments,  \nfailing to actively reduce hypothesis uncertainty. To compare these results, we col  \nlect human data on the task, which reveals a gap in inductive reasoning, particularly  \nfor more complex rules. Overall, ZENDOWORLD takes an important step toward  \nevaluating intelligent agents and identifies concrete avenues for improvement, par  \nticularly in domains like scientific discovery.  \nCode: [https://github.com/ml-research/ZendoWorld](https://github.com/ml-research/ZendoWorld)  \nData: [https://huggingface.co/datasets/ss567uhg/zendo-synthetic-data](https://huggingface.co/datasets/ss567uhg/zendo-synthetic-data)  \n1 Introduction  \nThe ability to form, test, and revise hypotheses in light of new evidence is a hallmark of human intelligence, reflecting the interplay between perception and inductive reasoning. Cognitive science has long emphasized this process as central to learning, showing how humans actively explore hypothesis spaces through informative experiments [10, 7, 24] . From a causal perspective, this process can be understood as learning not only from observations but from interventions that probe underlying mechanisms [26, 27] . With the growing interest in AI agents and AI scientists [16, 20, 38], structured benchmarks and controlled analyses are needed to assess these capabilities systematically.  \nRecent progress in AI has focused on inductive reasoning from few-shot examples [31, 19, 12, 37], typically evalu-  \nFigure 1: ZENDOWORLD combines perception, induction, and experimentation in a controlled visual environment.  \nated on grid-based puzzles such as ARC-AGI [6] or visual reasoning tasks over synthetic and real images [3, 39, 14, 34, 30] (cf. Figure 1 blue ∩ yellow regions) . While these benchmarks highlight  \nPreprint.  \nTable 1: Taxonomy of benchmarks at the intersection of perception, reasoning and experimentation. Our environment (ZENDOWORLD) is the first to require grounded visual induction through active experimentation.  \n\n| Benchmark / Env | Primary Modality | Perception | Induction | Experimentation |\n| --- | --- | --- | --- | --- |\n| ARC-AGI-1/2 [6] | Symbolic (Grid) | × | \u003Cbr>✓ | × |\n| Bongard, RAVEN [3, 39, 23] | Graphic | ✓ | ✓ | × |\n| Bongard-style [14, 34, 21, 25] | Visual (3D) | ✓ | ✓ | × |\n| Atari, NetHack [17, 33] | Visual (2D) | \u003Cbr>✓ | ×* | \u003Cbr>✓ |\n| MineDojo [9] | Visual (3D) | ✓ | ×* | ✓ |\n| CLEVR-AVR [40] | Visual (3D) | ✓ | ×** | ✓ |\n| Zendo, ActiveACRE [4][28] | Symbolic (Text) | × | \u003Cbr>✓ | \u003Cbr>✓ |\n| ARC-AGI-3, AutumnBench [1, 32] | Symbolic (Gr","cbCaiumQ7DF9Fmhp","https://ap.wps.com/l/cbCaiumQ7DF9Fmhp","pdf",2610758,1,33,"English","en",105,"# Introduction\n## Related Work and Motivation\n## ZendoWorld Benchmark\n# Agent Evaluation Approach\n## Agent Variants and Methods\n# Key Findings and Human Comparison","[{\"question\":\"How do different agent types struggle in ZendoWorld?\",\"answer\":\"The results suggest distinct bottlenecks: perception and induction limit different agent classes, and VLM-based agents tend to propose near-uninformative experiments that fail to actively reduce hypothesis uncertainty.\"}]",1784186929,83,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":78,"head_meta":80,"extra_data":82,"updated_unix":27},"playing-zendoworld-challenging-ai-agents-on-active-visual-concept-induction","",{"@graph":35,"@context":77},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/playing-zendoworld-challenging-ai-agents-on-active-visual-concept-induction/83348/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71],{"name":72,"@type":73,"acceptedAnswer":74},"How do different agent types struggle in ZendoWorld?","Question",{"text":75,"@type":76},"The results suggest distinct bottlenecks: perception and induction limit different agent classes, and VLM-based agents tend to propose near-uninformative experiments that fail to actively reduce hypothesis uncertainty.","Answer","https://schema.org",{"og:url":51,"og:type":79,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":81,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":84},[85,89,93,97,102,107,112,115,120,123,127],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":86,"show_sort_weight":87,"slug":88},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":90,"show_sort_weight":91,"slug":92},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Exam",70,"exam",{"id":98,"doc_module":4,"doc_module_name":45,"category_name":99,"show_sort_weight":100,"slug":101},5,"Comic",60,"comic",{"id":103,"doc_module":4,"doc_module_name":45,"category_name":104,"show_sort_weight":105,"slug":106},6,"Technology",50,"technology",{"id":108,"doc_module":4,"doc_module_name":45,"category_name":109,"show_sort_weight":110,"slug":111},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":113,"slug":114},30,"research-report",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":45,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":98,"slug":130},19,"General","general"]