Top AI researchers look beyond LLMs

AI Technology & Innovation - Tencent's 3D world generation model is now industry-leading, signifying a push into AI beyond language models [1] - World models represent a new frontier in AI, attracting major companies and researchers [2] - World models simulate reality, understanding depth, physics, and context from single images, unlike LLMs that describe reality [4] - World models may not require the same brute force GPU power as LLMs, applying learned rules instead of crunching probabilities [5] - Spatial intelligence, teaching machines to understand the world like humans, is a key concept behind world models [6] Market Potential & Applications - World models could open up entirely new markets and change the economics of AI [7] - Potential use cases include real estate (touring houses), interior design, digital twins, manufacturing, and autonomous driving [8][9] Competitive Landscape - Companies like Tencent, Google, Nvidia, and World Labs are actively involved in developing world models [2] - Meta's chief AI scientist is reportedly leaving to focus on world models [2] Cost & Efficiency - A basic 3D model can be generated quickly (within 5 minutes) from a single image, with options for more detailed models at a cost (e.g., $20 for more photos) [3][7][8]