Metraipolis › archive › dc_fa4283a2a34e
Do World Models Learn Global Understanding? — arXiv:2609.34058
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.34058
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T21:02:03.404Z
- SHA-256 of the archived text
959b0e75f0e84093617cea6ed46163d8145202e8d6049b3c5e9b8e59906aa081- Stable URL of this page
https://metraipolis.com/documento/dc_fa4283a2a34e
How to cite
Do World Models Learn Global Understanding? — arXiv:2609.34058 — arXiv (en), https://arxiv.org/abs/2609.34058 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fa4283a2a34e (Metraipolis archive dc_fa4283a2a34e, sha256 959b0e75f0e84093617cea6ed46163d8145202e8d6049b3c5e9b8e59906aa081)
BibTeX
@misc{dcfa4283a2a34e,
title = {Do World Models Learn Global Understanding? — arXiv:2609.34058},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.34058},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fa4283a2a34e; sha256 959b0e75f0e84093617cea6ed46163d8145202e8d6049b3c5e9b8e59906aa081},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34058. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
AI systems often feel brittle and fragmented. A large language model (LLM) may correctly explain a concept but fail to apply it, or follow safety instructions in one context but not another. This behavior suggests a general failure to lift local information to a global understanding. To gain fundamental insight, we frame "understanding" as learning constraints and propagating their consequences. We construct learning tasks on monoid worlds, sets of states connected by action transitions, where observed training transitions and an unseen constraint jointly determine held-out transitions. Measuring generalization tests whether models can learn global constraints from local transitions and propagate their consequences. We consider inverse, commutativity, composition, and periodicity constraints relevant to spatial and semantic structure. Across attention, recurrent, and state-space architectures, next-state training fits the data but fails to propagate non-trivial constraints. Compositional training, which uses identical paths but hides intermediate states from the input, achieves 96% accuracy on inverse, commutativity, and composition constraints across architectures, yields corresponding improvements in geometric generalization of world models trained on embodied environments and relational generalization in Wikidata-finetuned LLMs. How far do models propagate constraints when inferring an unseen fact may depend on first inferring others? We define proof depth d of a held-out transition, measuring the minimum number of inference rounds to infer the transition, and find that model generalization decreases sharply with proof depth. Increasing compositional path length T improves generalization. These results provide a formal way to investigate global understanding in language and world models and demonstrate that compositional training promotes information propagation and integration. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.34058 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T20:29:02.832Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).