LLM, Quantum AI, and World Models: A Revolution in Progress

Large language models (LLMs) are at the heart of current advancements in artificial intelligence, but their understanding of the real world remains limited. To bridge this gap, many startups are focusing on the development of "world models." These models aim to provide AI with a more human and grounded understanding of causality, inspired by cognitive neuroscience.

Influential figures such as Yann LeCun and Fei-Fei Li, along with tech giants like Google, Meta, Tencent, and Nvidia, are investing in this direction. Companies like Physical Intelligence are developing foundational models for physical AI, enabling robots to learn and make decisions more effectively. Packy McCormick is also mentioned among those highlighting the limitations of current LLMs.

James Wang, a guest contributor and VC in AI and DeepTech, is the author of the newsletter "Weighty Thoughts." His book "What You Need to Know About AI" recently won the American Legacy Award in the Best Non-Fiction category.

Meanwhile, quantum computing is making significant strides. Nvidia recently announced Nvidia Ising, an open AI model that could accelerate the achievement of useful quantum computing. This model is being tested by renowned institutions such as Academia Sinica, Fermi National Accelerator Laboratory, and Harvard. A recent study conducted by researchers from Caltech, Google Quantum AI, MIT, and Oratomic demonstrates that small quantum systems can efficiently process large amounts of data, thereby reducing memory requirements.

In the field of robotics, Google has launched Gemini Robotics-ER 1.6, which enhances the visual and spatial understanding of robots. This system allows for better object identification and strengthens robotic safety by integrating knowledge of the physical world, such as avoiding liquids or objects weighing over 20 kg.

However, the development of world models presents challenges, particularly due to data friction. Despite this, investment in physical AI is booming, with companies like World Labs, Skild AI, and AMI Labs raising billions of dollars to overcome these obstacles. AMI Labs has raised $1.03 billion by early 2026.