⌁ Awesome World Model Hub Papers · Datasets · Projects
中文 GitHub
← Back to papers

Quo Vadis, World Modeling?

arXiv 2026 59.3 theory, survey

TLDR

Proposes agent-centric interactive world proxies, shifting from physical state prediction to broader feedback modalities for agent improvement.

Reasoning

The paper offers a novel conceptual framework by categorizing world proxies into six functional forms and three agent empowerment levels, which is a strength. However, it lacks empirical validation or real-world experiments, making it purely theoretical.

Read-first score

Read-first score 59.3, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 31.

Methodology quality 25%
100

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=benchmark,evaluation,experiment,result,validation

Recency 8%
100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Topical relevance 42%
44.3

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Reproducibility 25%
30

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Field roles

FrontierMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 480.

Keyword Scores

world model
9
interactive world model
8
world simulator
5
generative world model
3
world dynamics prediction
3
model-based reinforcement learning world model
2
video world model
1

Deep Analysis

Innovations

  • Shifts world modeling from physical state prediction to agent-usable information transitions, introducing Agent-Centric Interactive World Proxies.
  • Organizes world proxies into six functional forms: dynamics, spatial, execution, memory/experience, skill, and reward/verification proxies.
  • Defines three progressive levels of agent-proxy interaction: Inference-Time Guidance, Training-Time Optimization, and Agent-Proxy Co-Evolution.

Methodology

This is a conceptual paper that proposes a taxonomy and design space. It systematically maps agent-centric world proxies by categorizing feedback modalities into six functional forms and analyzing how they empower agents across three levels of increasing autonomy and adaptation.

Key Results

No empirical results are reported; the paper presents a conceptual framework and roadmap for building world proxies that support continual agent improvement.

Limitations

  • Purely conceptual with no experimental validation or empirical evidence.
  • Relies on proposed taxonomies and levels without quantitative benchmarks or case studies.
  • Lacks concrete implementation details or evaluation of the six proxy forms and three levels.

Resources

Paper

Tags