Alexandre LeBrun, CEO of AMI Labs, has opted not to label his company’s AI work as 'AGI' or 'superintelligence,' a stance that sets him apart in an industry eager to adopt such terms. During an interview with TechCrunch, LeBrun emphasized that the company does not use these labels at all, noting that the terminology has shifted from 'AGI' to 'superintelligence' and may change again in the future. He questioned the usefulness of the term 'superintelligence,' stating, 'There’s no good definition. What is superintelligence? I don’t know. It’s not a very useful word.'
LeBrun, who is based in Seoul for the International Conference on Machine Learning, explained that AMI Labs is still pre-product but is actively seeking partnerships in robotics, manufacturing, and electronics. He highlighted the need for world models to prove their value beyond the lab, particularly in robotics, where current systems are limited to fixed routines. 'Robots are not safe right now,' he said. 'There’s no solution for that today.' The goal, he said, is to develop context-aware AI that can prevent accidents, such as a robot mistaking a child for a dance partner during a public event.
LeBrun also discussed the complementary roles of large language models (LLMs) and world models, arguing that the latter are essential for real-world understanding. He pointed to the limitations of LLMs in physical environments, suggesting that world models are necessary for industries that interact with the real world, such as manufacturing. Despite the company’s lack of a product yet, LeBrun emphasized the importance of real-world data and partnerships, particularly in Asia, where AMI is exploring opportunities.
Source: techcrunch
