Home / Technology / AI CEO Rejects 'AGI' Hype, Focuses on 'World Models'
AI CEO Rejects 'AGI' Hype, Focuses on 'World Models'
16 Jul
Summary
- Startup CEO avoids 'AGI' and 'superintelligence' labels for AI.
- World models predict next states, unlike LLMs predicting next words.
- Company seeks partners in Asia for real-world AI model training.
Alexandre LeBrun, CEO of AMI Labs, co-founded by AI pioneer Yann LeCun, is deliberately avoiding labels like "AGI" and "superintelligence" to describe his company's AI advancements. LeBrun expressed skepticism regarding the definitions of these terms, stating they are not particularly useful. Instead, AMI Labs is focusing on developing "world models" designed to predict the next state of the physical world, a significant departure from Large Language Models (LLMs) that predict text.
AMI Labs, still pre-product, is actively pursuing partnerships in robotics, manufacturing, and electronics. LeBrun highlighted the current limitations of AI in the physical world, where robots often operate on fixed routines. He explained that world models aim to provide robots with context awareness, enabling safer and more adaptable operations. These models are seen as complementary to LLMs, which will continue to excel at language processing.
The company's strategy involves training these world models in real-world environments, leading LeBrun to focus on Asia, particularly South Korea. He cited South Korea's advanced robotics, semiconductor, and manufacturing sectors, as well as its rapid adoption of new technologies, as key attractions for AMI Labs' development and partnership goals. The company aims to be at the forefront of this physical AI revolution.
