How Should World Models Be Evaluated? A Decision-Making-Centric Position

• URL: http://arxiv.org/abs/2606.15032v1
• Date: Sat, 13 Jun 2026 00:21:21 GMT
• Title: How Should World Models Be Evaluated? A Decision-Making-Centric Position
• Abstract summary: World models have rapidly become one of the central abstractions in modern AI.<n>Recent papers measure video realism, perceptual similarity, instruction following, physical plausibility, policy ranking, executability, planning success, and downstream policy improvement.<n>This paper argues that the central question is use-dependent.
• Score: 5.773605711982218
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: World models have rapidly become one of the central abstractions in modern AI. Yet the term now refers to several different objects: action-conditioned environment models, latent imagination models, future-video predictors, interactive neural simulators, latent predictive representations, and synthetic-data engines. Evaluation has broadened with the term. Recent papers measure video realism, perceptual similarity, instruction following, physical plausibility, policy ranking, executability, planning success, and downstream policy improvement. The result is not only metric diversity but also a recurring problem of claim/evidence mismatch: papers frequently make a stronger claim about what their model is useful for than their evaluation can actually establish. This paper surveys the recent literature and argues that the central question is use-dependent. When a model is presented as a world model for embodied decision-making, a more decisive issue is not whether it generates visually compelling videos, but whether it supports reliable counterfactual reasoning, policy evaluation, planning, and policy optimization under intervention, policy-induced distribution shift, and long-horizon rollout. We organize the literature using an L0--L7 ladder that ranges from visual plausibility to policy optimization utility. In our interpretation, L0--L3 are most naturally read as diagnostics of generated artifacts, L4 is often the first genuinely interventional test, and L5--L7 provide the most direct evidence of decision usefulness. Based on this diagnosis, we propose a decision-making-centric evaluation framework and a benchmark protocol that foreground counterfactual action fidelity, closed-loop rollout validity, reward/value prediction, policy-ranking agreement, optimization lift, model exploitability, and uncertainty calibration.

Related papers

• SPHINX: First Explain, Then Explore [9.461724985210312]
  We argue that adversarial scenes should be generated based on the failure diagnosis of the driving policy.<n>In this paper, we propose a closed-loop framework for adversarial scenario synthesis guided by a simple principle: first explain, then explore.
  arXiv  Detail & Related papers  (2026-06-16T03:48:27Z)
• Can Revealed Preferences Clarify LLM Alignment and Steering? [23.175180848107093]
  We present an empirical pipeline for estimating the implied preferences that an LLM's observed choices optimize.<n>We show how this revealed-preference description allows rigorous evaluation of whether models behave in a consistently goal-directed way.
  arXiv  Detail & Related papers  (2026-05-08T23:26:35Z)
• Agentic World Modeling: Foundations, Capabilities, Laws, and Beyond [209.35045331678043]
  We introduce a "levels x laws" taxonomy organized along two axes.<n>The first defines three capability levels: L1 Predictor, which learns one-step local transition operators; L2 Simulator, which composes them into multi-step, action-conditioned rollouts that respect domain laws; and L3 Evolver, which autonomously revises its own model when predictions fail against new evidence.<n>We synthesize over 400 works and summarize more than 100 representative systems spanning model-based reinforcement learning, video generation, web and GUI agents, multi-agent social simulation, and AI-driven scientific discovery.
  arXiv  Detail & Related papers  (2026-04-24T17:48:47Z)
• MANTA: Multi-turn Assessment for Nonhuman Thinking & Alignment [0.1452394725421793]
  MANTA is a dynamic multi-turn evaluation framework built on the Inspect AI platform.<n>We present preliminary results from evaluations of claude-sonnet-4-20250514 and openai/gpt-4o.<n>We additionally present STYLEJUDGE, a controlled four-judge study demonstrating systematic format bias in LLM-as-judge evaluation.
  arXiv  Detail & Related papers  (2026-04-18T19:51:32Z)
• World Action Verifier: Self-Improving World Models via Forward-Inverse Asymmetry [82.93104394404781]
  General-purpose world models promise scalable policy evaluation, optimization, and planning.<n>We propose World Action Verifier (WAV), a framework that enables world models to identify their own prediction errors and self-improve.
  arXiv  Detail & Related papers  (2026-04-02T12:48:36Z)
• What-If Analysis of Large Language Models: Explore the Game World Using Proactive Thinking [50.72154186522052]
  Large language models (LLMs) excel at processing information reactively but lack the ability to systemically explore hypothetical futures.<n>We propose WiA-LLM, a new paradigm that equips LLMs with proactive thinking capabilities.<n>We validate WiA-LLM in Honor of Kings, a complex multiplayer game environment.
  arXiv  Detail & Related papers  (2025-09-05T04:05:27Z)
• Large Language Models Reasoning Abilities Under Non-Ideal Conditions After RL-Fine-Tuning [33.27410930782468]
  We introduce a new research direction guided by brain-science findings that human reasoning remains reliable under imperfect inputs.<n>We fine-tune three large language models (LLMs) and a state-of-the-art large vision-language model (LVLM) usingReinforcement learning (RL) with a representative policy-gradient algorithm.<n>Our results reveal that while RL fine-tuning improves baseline reasoning under idealized settings, performance declines significantly across all three non-ideal scenarios.
  arXiv  Detail & Related papers  (2025-08-06T19:51:29Z)
• Wait, that's not an option: LLMs Robustness with Incorrect Multiple-Choice Options [2.1184929769291294]
  This work introduces a novel framework for evaluating LLMs' capacity to balance instruction-following with critical reasoning.<n>We show that post-training aligned models often default to selecting invalid options, while base models exhibit improved refusal capabilities that scale with model size.<n>We additionally conduct a parallel human study showing similar instruction-following biases, with implications for how these biases may propagate through human feedback datasets used in alignment.
  arXiv  Detail & Related papers  (2024-08-27T19:27:43Z)
• Belief Revision: The Adaptability of Large Language Models Reasoning [63.0281286287648]
  We introduce Belief-R, a new dataset designed to test LMs' belief revision ability when presented with new evidence. Inspired by how humans suppress prior inferences, this task assesses LMs within the newly proposed delta reasoning framework. We evaluate $sim$30 LMs across diverse prompting strategies and found that LMs generally struggle to appropriately revise their beliefs in response to new information.
  arXiv  Detail & Related papers  (2024-06-28T09:09:36Z)
• Dual policy as self-model for planning [71.73710074424511]
  We refer to the model used to simulate one's decisions as the agent's self-model. Inspired by current reinforcement learning approaches and neuroscience, we explore the benefits and limitations of using a distilled policy network as the self-model.
  arXiv  Detail & Related papers  (2023-06-07T13:58:45Z)
• Mismatched No More: Joint Model-Policy Optimization for Model-Based RL [172.37829823752364]
  We propose a single objective for jointly training the model and the policy, such that updates to either component increases a lower bound on expected return. Our objective is a global lower bound on expected return, and this bound becomes tight under certain assumptions. The resulting algorithm (MnM) is conceptually similar to a GAN.
  arXiv  Detail & Related papers  (2021-10-06T13:43:27Z)
• Counterfactual Learning of Stochastic Policies with Continuous Actions [42.903292639112536]
  We introduce a modelling strategy based on a joint kernel embedding of contexts and actions.<n>We empirically show that the optimization aspect of counterfactual learning is important.<n>We propose an evaluation protocol for offline policies in real-world logged systems.
  arXiv  Detail & Related papers  (2020-04-22T07:42:30Z)

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.