# Playing ZendoWorld: Challenging AI Agents on Active Visual Concept Induction

arXiv: https://arxiv.org/abs/2607.08233 · HTML: https://mlsanity.com/paper/2607.08233/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal and embodied agents, one of 16 directions in LLM agents and tool use.
- Builds on 25 of its 40 references that are in the map.
- First posted on arXiv 9 Jul 2026.

## Placed in

- LLM agents and tool use / [Multimodal and embodied agents](https://mlsanity.com/agents/multimodal-and-embodied-agents/)

## Builds on (most cited first, 25 of 25)

- [The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery](https://mlsanity.com/paper/2408.06292/) (2024)
- [MineDojo: Building Open-Ended Embodied Agents with Internet-Scale Knowledge](https://mlsanity.com/paper/2206.08853/) (2022)
- [RAVEN: A Dataset for Relational and Analogical Visual rEasoNing](https://mlsanity.com/paper/1903.02741/) (2019)
- [The NetHack Learning Environment](https://mlsanity.com/paper/2006.13760/) (2020)
- [DreamCoder: Growing generalizable, interpretable knowledge with wake-sleep Bayesian program learning](https://mlsanity.com/paper/2006.08381/) (2020)
- [Hypothesis Search: Inductive Reasoning with Language Models](https://mlsanity.com/paper/2309.05660/) (2023)
- [Bongard-LOGO: A New Benchmark for Human-Level Concept Learning and Reasoning](https://mlsanity.com/paper/2010.00763/) (2020)
- [Combining Induction and Transduction for Abstract Reasoning](https://mlsanity.com/paper/2411.02272/) (2024)
- [Visual Agentic AI for Spatial Reasoning with a Dynamic API](https://mlsanity.com/paper/2502.06787/) (2025)
- [ARC-AGI-3: A New Challenge for Frontier Agentic Intelligence](https://mlsanity.com/paper/2603.24621/) (2026)
- [Bongard-HOI: Benchmarking Few-Shot Visual Reasoning for Human-Object Interactions](https://mlsanity.com/paper/2205.13803/) (2022)
- [Doing Experiments and Revising Rules with Natural Language and Probabilistic Reasoning](https://mlsanity.com/paper/2402.06025/) (2024)
- [Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World](https://mlsanity.com/paper/2310.10207/) (2023)
- [Pix2Code: Learning to Compose Neural Visual Concepts as Programs](https://mlsanity.com/paper/2402.08280/) (2024)
- [Bongard in Wonderland: Visual Puzzles that Still Make AI Go Mad?](https://mlsanity.com/paper/2410.19546/) (2024)
- [Reasoning Limitations of Multimodal Large Language Models. A Case Study of Bongard Problems](https://mlsanity.com/paper/2411.01173/) (2024)
- [How Far Are AI Scientists from Changing the World?](https://mlsanity.com/paper/2507.23276/) (2025)
- [Object Centric Concept Bottlenecks](https://mlsanity.com/paper/2505.24492/) (2025)
- [Do AI Models Perform Human-like Abstract Reasoning Across Modalities?](https://mlsanity.com/paper/2510.02125/) (2025)
- [Atari-GPT: Benchmarking Multimodal Large Language Models as Low-Level Policies in Atari Games](https://mlsanity.com/paper/2408.15950/) (2024)
- [Bongard-RWR+: Real-World Representations of Fine-Grained Concepts in Bongard Problems](https://mlsanity.com/paper/2508.12026/) (2025)
- [Scaling Neural Program Synthesis with Distribution-based Search](https://mlsanity.com/paper/2110.12485/) (2021)
- [SLR: Automated Synthesis for Scalable Logical Reasoning](https://mlsanity.com/paper/2506.15787/) (2025)
- [Synthesizing Visual Concepts as Vision-Language Programs](https://mlsanity.com/paper/2511.18964/) (2025)
- [PhysVLM-AVR: Active Visual Reasoning for Multimodal Large Language Models in Physical Environments](https://mlsanity.com/paper/2510.21111/) (2025)

## Built on by (most cited first, 1 of 1)

- [Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models](https://mlsanity.com/paper/2608.09696/) (2026)

## Abstract (arXiv)

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.