Review – 2
WSJ: “Today’s AIs are book smart. Everything they know they learned from available language, images and videos. To evolve further, they have to get street smart. That requires “world models.” The key is enabling AI to learn from their environments and faithfully represent an abstract version of them in their “heads,” the way humans and animals do. To do it, developers need to train AIs by using simulations of the world… There’s an almost unanimous belief among AI pioneers that world models are crucial to creating next-generation AI. And many say they will be critical to someday creating better-than-human “artificial general intelligence,” or AGI… While the type of AI that makes large language models and ChatGPT possible gets all of the attention right now, it’s world-model-based AI that is gaining momentum in frontier research and could allow technology to take on new roles in our lives… LLMs appear to be taking over some functions in white-collar jobs already, and AIs that are world-model smart could allow them to take over yet more. Blue-collar work has been relatively safe. But as AI developers go ever deeper on world models, robots could start applying for jobs as truck drivers, plumbers or caregivers.”
WSJ: “Yann LeCun has been telling anyone who asks that he thinks large language models, or LLMs, are a dead end in the pursuit of computers that can truly outthink humans… “We are not going to get to human-level AI just by scaling LLMs,” he said.”
Fei-Fei Li: “Today, leading AI technology such as large language models (LLMs) have begun to transform how we access and work with abstract knowledge. Yet they remain wordsmiths in the dark; eloquent but inexperienced, knowledgeable but ungrounded. Spatial intelligence will transform how we create and interact with real and virtual worlds—revolutionizing storytelling, creativity, robotics, scientific discovery, and beyond. This is AI’s next frontier…Building spatially intelligent AI requires something even more ambitious than LLMs: world models, a new type of generative models whose capabilities of understanding, reasoning, generation and interaction with the semantically, physically, geometrically and dynamically complex worlds – virtual or real – are far beyond the reach of today’s LLMs… For spatial intelligence, I define world models through three essential capabilities: 1. Generative: World models can generate worlds with perceptual, geometrical, and physical consistency…2. Multimodal: World models are multimodal by design…3. Interactive: World models can output the next states based on input actions.”
Daniel Ince-Cushman: “In the evolving landscape of artificial intelligence, two prominent paradigms have emerged although World Models remain lesser known. One is the Large Language Model (LLM), a linguistic construct trained to predict and generate human-like text. The other is the World Model, a computational framework focused on simulating environments and enabling decision-making through internalized representations of external dynamics…World models…are designed to learn representations of environments, often through multi-modal sensory data or interactive simulation. These models typically consist of perception modules (e.g., vision encoders), memory or state encoders, and predictive components that simulate future states given current inputs and actions. These systems embody fundamentally distinct conceptions of intelligence. The LLM derives its power from statistical patterns in textual data, while the world model seeks to emulate causality and the laws of physics through learned simulations. The utility of a world model lies in its ability to enable planning. Through internal simulation, an agent can forecast outcomes and select optimal actions without direct environmental interaction.”
Gilbert Traverse: “The most valuable enterprise AI investments going forward will likely focus on integrating these two paradigms: leveraging LLMs for human-centric interfaces and general knowledge, while deploying increasingly sophisticated World Models to drive the complex planning, optimization, and action needed for real-world automation in areas like manufacturing, logistics, and autonomous operations. Strategic foresight demands an understanding of both, allowing organizations to capitalize on the linguistic power of today’s AI while building the causal foundation for tomorrow’s intelligent agents.”
