Playing ZendoWorld: Challenging AI Agents on Active Visual Concept Induction

• URL: http://arxiv.org/abs/2607.08233v1
• Date: Thu, 09 Jul 2026 08:29:23 GMT
• Title: Playing ZendoWorld: Challenging AI Agents on Active Visual Concept Induction
• Abstract summary: ZendoWorld is a controlled interactive environment in which agents must infer a logical rule about visual game observations.<n>We evaluate several agents spanning pure VLM reasoning, Bayesian particle filtering, dynamic concept discovery, and neuro-symbolic methods.<n>Our main findings are: (1) high accuracy in predicting labels does not imply recovery of the underlying rule; (2) perception and induction are distinct bottlenecks for different agent classes; and (3) VLM-based agents propose near-uninformative experiments, failing to actively reduce hypothesis uncertainty.
• Score: 39.71864879537297
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: A central challenge in building intelligent systems is enabling agents to jointly perceive complex inputs, form hypotheses about hidden patterns, and design informative experiments to test them. To study this problem, we propose ZendoWorld, a controlled interactive environment in which agents must infer a logical rule about visual game observations, acquire information by proposing new scenes, and refine their hypotheses based on feedback from the game environment. We evaluate several agents spanning pure VLM reasoning, Bayesian particle filtering, dynamic concept discovery, and neuro-symbolic methods. Our main findings are: (1) high accuracy in predicting labels for observed examples does not imply recovery of the underlying rule; (2) perception and induction are distinct bottlenecks for different agent classes; and (3) VLM-based agents propose near-uninformative experiments, failing to actively reduce hypothesis uncertainty. To compare these results, we collect human data on the task, which reveals a gap in inductive reasoning, particularly for more complex rules. Overall, ZENDOWORLD takes an important step toward evaluating intelligent agents and identifies concrete avenues for improvement, particularly in domains like scientific discovery.

Related papers

• Unleashing Multimodal Large Language Models for Training-free HOI Detection in the Wild [49.77540427384148]
  We present AgentHOI, a training-free, agentic framework that transfers the generalist multimodal reasoning capabilities of foundation models to HOI detection in the wild.<n>To address the challenges of incomplete interaction discovery and ambiguous localization in complex scenes, we introduce two key mechanisms.<n>AgentHOI achieves superior performance over state-of-the-art supervised and weakly supervised methods in real-world settings.
  arXiv  Detail & Related papers  (2026-07-15T14:30:20Z)
• ESI-Bench: Towards Embodied Spatial Intelligence that Closes the Perception-Action Loop [55.468404995694975]
  We develop a benchmark for embodied spatial intelligence spanning 10 task categories and 29 subcategories built on OmniGibson.<n>We conduct experiments on state-of-the-art MLLMs and find that active exploration substantially outperforms passive counterparts.<n>Unlike humans who seek falsifying viewpoints and revise beliefs under contradiction, models commit prematurely with high confidence regardless of evidence quality.
  arXiv  Detail & Related papers  (2026-05-18T17:59:02Z)
• What You Think is What You See: Driving Exploration in VLM Agents via Visual-Linguistic Curiosity [38.38728887407681]
  We ask: Can VLM agents actively find signals that challenge and refine their internal world model through curiosity-driven exploration?<n>We propose GLANCE, a unified framework that bridges reasoning and exploration by grounding the agent's linguistic world model into the stable visual representations of an evolving target network.
  arXiv  Detail & Related papers  (2026-05-05T14:08:54Z)
• Theory of Space: Can Foundation Models Construct Spatial Beliefs through Active Exploration? [83.13508919229939]
  Theory of Space is defined as an agent's ability to actively acquire information through self-directed, active exploration.<n>A key innovation is spatial belief probing, which prompts models to reveal their internal spatial representations at each step.<n>Our findings suggest that current foundation models struggle to maintain coherent, revisable spatial beliefs during active exploration.
  arXiv  Detail & Related papers  (2026-02-04T19:06:40Z)
• DISCOVERYWORLD: A Virtual Environment for Developing and Evaluating Automated Scientific Discovery Agents [49.74065769505137]
  We introduce DISCOVERYWORLD, the first virtual environment for developing and benchmarking an agent's ability to perform complete cycles of novel scientific discovery. It includes 120 different challenge tasks spanning eight topics each with three levels of difficulty and several parametric variations. We find that strong baseline agents, that perform well in prior published environments, struggle on most DISCOVERYWORLD tasks.
  arXiv  Detail & Related papers  (2024-06-10T20:08:44Z)
• Assessing biomedical knowledge robustness in large language models by query-efficient sampling attacks [0.6282171844772422]
  An increasing depth of parametric domain knowledge in large language models (LLMs) is fueling their rapid deployment in real-world applications.<n>The recent discovery of named entities as adversarial examples in natural language processing tasks raises questions about their potential impact on the knowledge robustness of pre-trained and finetuned LLMs.<n>We developed an embedding-space attack based on powerscaled distance-weighted sampling to assess the robustness of their biomedical knowledge.
  arXiv  Detail & Related papers  (2024-02-16T09:29:38Z)
• Interactive Visual Reasoning under Uncertainty [29.596555383319814]
  We devise the IVRE environment for evaluating artificial agents' reasoning ability under uncertainty. IVRE is an interactive environment featuring rich scenarios centered around Blicket detection.
  arXiv  Detail & Related papers  (2022-06-18T13:32:41Z)
• Information is Power: Intrinsic Control via Information Capture [110.3143711650806]
  We argue that a compact and general learning objective is to minimize the entropy of the agent's state visitation estimated using a latent state-space model. This objective induces an agent to both gather information about its environment, corresponding to reducing uncertainty, and to gain control over its environment, corresponding to reducing the unpredictability of future world states.
  arXiv  Detail & Related papers  (2021-12-07T18:50:42Z)
• On the Sensory Commutativity of Action Sequences for Embodied Agents [2.320417845168326]
  We study perception for embodied agents under the mathematical formalism of group theory. We introduce the Sensory Commutativity Probability criterion which measures how much an agent's degree of freedom affects the environment. We empirically illustrate how SCP and the commutative properties of action sequences can be used to learn about objects in the environment.
  arXiv  Detail & Related papers  (2020-02-13T16:58:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.