{"slug": "simulating-everything-sort-of-the-promise-and-limits-of-world-models", "title": "Simulating everything, sort of: The promise and limits of world models", "summary": "World models, a new category of AI systems designed to simulate the physical world, are gaining momentum as researchers and companies seek to overcome the limitations of large language models. Experts from MIT, Runway, and World Labs discuss how these models focus on applications in robotics, research, and asset generation, though their interfaces and tools remain undefined. Yann LeCun argues that extending LLMs to human-level intelligence is impossible, positioning world models as a more promising path forward.", "body_md": "Over the past few years, many of us have gotten a crash course in what we now call artificial intelligence—but really, it has mostly been a crash course in large language models. Increasingly, however, LLMs are no longer the only category of AI drawing high expectations, massive funding rounds, and significant research and product development.\n\nOver the past year, we’ve seen a plethora of new announcements in a category labeled “world models,” and you’ll likely see more movement there in the coming months and years.\n\nInstead of or in addition to working with language, world models aim to lay the groundwork for AI systems that are capable of simulating the physical world, or at least a useful approximation of it.\n\nTo examine what’s different and important about this idea, Ars spoke with three expert practitioners working on world models and related technologies: [Vincent Sitzmann](https://www.vincentsitzmann.com/) from MIT, [Anastasis Germanidis](https://agermanidis.com/) from Runway, and [Ben Mildenhall](https://bmild.github.io/) from World Labs.\n\nFrom these conversations, we learned that while LLMs-as-a-product started with an interface (chat) and then sought a use case, the big players in world models right now are arguably working in the other direction: They’re starting with specific use cases and applications in robotics, research, and asset generation, but it’s unclear exactly how the interfaces, systems, and tools will ultimately look.\n\n## The off-ramp from LLM disillusionment\n\nAs you’ll soon see, there are many parallels between LLMs and world models in terms of architecture and how people expect them to improve over time. For some, though, they’re seen as a potential answer to the limitations of LLMs, even though work on them predates that contemporary narrative.\n\n“The idea that you’re going to extend the capabilities of LLMs to the point that they’re going to have human-level intelligence is complete nonsense,” former Meta chief AI scientist Yann LeCun [told Wired](https://www.wired.com/story/yann-lecun-raises-dollar1-billion-to-build-ai-that-understands-the-physical-world/) earlier this year. LeCun has made waves with an opinion that some working in AI and LLMs see as contrarian, but he’s actually speaking for a sizable segment of the field.", "url": "https://wpnews.pro/news/simulating-everything-sort-of-the-promise-and-limits-of-world-models", "canonical_source": "https://arstechnica.com/ai/2026/07/simulating-everything-sort-of-the-promise-and-limits-of-world-models/", "published_at": "2026-07-13 11:00:51+00:00", "updated_at": "2026-07-13 16:06:49.646881+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "ai-research", "ai-products", "ai-startups"], "entities": ["MIT", "Runway", "World Labs", "Vincent Sitzmann", "Anastasis Germanidis", "Ben Mildenhall", "Yann LeCun"], "alternates": {"html": "https://wpnews.pro/news/simulating-everything-sort-of-the-promise-and-limits-of-world-models", "markdown": "https://wpnews.pro/news/simulating-everything-sort-of-the-promise-and-limits-of-world-models.md", "text": "https://wpnews.pro/news/simulating-everything-sort-of-the-promise-and-limits-of-world-models.txt", "jsonld": "https://wpnews.pro/news/simulating-everything-sort-of-the-promise-and-limits-of-world-models.jsonld"}}