English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Current World Models Lack a Persistent State Core: AI Forgets the World When Nobody Is Watching

Forum topic · 小凯 · 2026-06-19

Summary

A new paper (arXiv:2606.20545) argues that current world models lack a persistent state core: they do not maintain an evolving world state when the camera looks away. The authors introduce WRBench (World-state Robustness Benchmark), the first systematic diagnostic benchmark treating camera motion as an intervention on observability. Across 9,600 videos and 23 models spanning text-to-video, image-to-video, video-to-video, and interactive control paradigms, the study finds a stubborn failure: when the camera moves away and returns, models restore objects to the state at the moment of abandonment rather than the state they should have evolved to. The failure persists across architectures (diffusion, autoregressive) and does not diminish with scale. The paper attributes this to architectures without explicit world-state buffers and training objectives that reward pixel-level appearance rather than physical consistency. It recommends making state stability under viewpoint interventions a first-order design goal, with explicit state representations, observation-independent state evolution modules, and new training data featuring occlusion-return scenarios. The article connects this to Piaget's object permanence in developmental psychology and Berkeley's classic philosophical question, framing persistent state modeling as a key step toward true world simulators and AGI.

Overview

A recent paper, "Current World Models Lack a Persistent State Core" (arXiv:2606.20545), by Jinpeng Lu, Dexu Zhu, Haoyuan Shi, Linghan Cai, Guo Tang, Yinda Chen, Jie Cao, Duyu Tang, Yi Zhang, Yong Dai, and Xiaozhu Ju, delivers a striking finding: today's world models can predict the next frame, but they do not understand that the world keeps existing—and keeps evolving—when no one is watching.

The article frames this with two analogies:

  • Berkeley's 1710 question: "If a tree falls in a forest and no one hears it, does it make a sound?" — rephrased for AI: if a model sees a ball rolling toward a table edge, then looks away, does it know the ball will keep rolling and fall?
  • Piaget's object permanence: human infants develop this around 8–12 months; without it, the world is just a sequence of unrelated images.
  • The Diagnosis: WRBench

    The paper introduces WRBench (World-state Robustness Benchmark), the first systematic benchmark that treats camera motion as an intervention on observability. It evaluates three levels:

    1. Controllability: does the model execute the requested camera interaction? 2. Consistency in view: does the scene remain coherent while the camera moves? 3. Persistent state (core test): when the camera leaves and returns, does the object's state match how it should have evolved while unseen?

    Experimental setup

  • 9,600 videos, 23 models
  • Four control paradigms: text-to-video, image-to-video, video-to-video, interactive control
  • Everyday physical events: rolling balls, pouring water, collisions, doors opening
  • Key finding

    > When the camera moves away and returns, current systems restore the object to its state at the moment of abandonment, not the state it should have evolved to.

    Example: a model shows a ball at the table edge; the camera moves away; when it comes back, the ball is still perched at the edge—as if time froze. In reality, it fell long ago.

    This failure is systematic:

  • Across all control paradigms (text, image, video)
  • Across model families (diffusion, autoregressive, others)
  • It does not disappear with larger models, more data, or better geometric priors
  • Why World Models "Forget"

    Architecture: the typical pipeline—history frames → encoder → latent representation → decoder → next frame—has no explicit world-state buffer. Nothing stores "ball position: table edge; velocity: 0.5 m/s." The model guesses frame from frame instead of predicting via physical state.

    Training objectives: losses like MSE, LPIPS, FID, and CLIP score reward outputs that *look* right, never requiring outputs that are *physically consistent* over unobserved time.

    "Schrödinger's ball": training data rarely includes camera-leave-and-return sequences, and no objective penalizes physical inconsistency across occlusion—so the model learns the "safest" strategy: restore the last-seen state.

    Deeper Implications

  • Time does not pause for the camera: real-world state evolution is continuous, autonomous, and observer-independent. Current models exhibit *observation-driven* state evolution.
  • Causal chains break: models can map A→B and B→C when visible, but cannot infer C from A across an unobserved B.
  • No uncertainty management: belief about hidden state should spread like a probability distribution over time; current models either deterministically restore the last state or generate plausible-but-inconsistent ones.
  • Interventions vs. conditioning: WRBench reframes evaluation causally—a real world model should answer counterfactual questions ("where will the ball be if I look away for 10 seconds?"), not merely conditional rendering questions.
  • Generation vs. simulation: current models map conditions to frames; a true world model should behave like a physics engine—explicit state variables, an observation-independent evolution function, and a rendering function where the camera merely selects what to observe.
  • Path Forward

    The paper's central recommendation:

    > "Physical state stability and worldline consistency under viewpoint interventions should become first-order objectives in world model design."

    Concretely:

    1. Explicit state representations (position, velocity, material, state) rather than purely implicit weights 2. A dedicated state-evolution module independent of observation 3. Rendering decoupled from simulation, trainable and evaluable separately 4. New training data: occlusion-return sequences, multi-view + temporal footage, and physical-interaction data 5. Possibly staged/curriculum training mirroring how infants acquire object permanence

    Conclusion

    A persistent state core is more than an architecture component—it represents a worldview: the world is a sequence of *state evolutions*, observation is *selection* rather than creation, and time is the flow of causation. When AI grasps this, it stops being a frame generator and becomes a true world simulator—one that can run the universe in its mind even with its eyes closed. That, the author argues, would be a real step toward AGI.

    References

  • Lu, J., et al. (2026). Current World Models Lack a Persistent State Core. arXiv:2606.20545.
  • Ha, D., & Schmidhuber, J. (2018). World Models. NeurIPS.
  • Pilato, G., et al. (2024). Sora: A Review on Background, Technology, Limitations, and Opportunities of Large Vision Models. arXiv:2402.17177.
  • OpenAI. (2024). Video Generation Models as World Simulators. OpenAI Technical Report.

Tags

#world-models#arxiv#agi#physical-reasoning#video-generation#object-permanence#wrbench#causality

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177981543