AMI Labs' AGI Dilemma: Founder Alexandre LeBrun Rejects 'Superintelligence' Term

Alexandre LeBrun, CEO of Yann LeCun's new AI startup AMI Labs, is refusing to join the industry's 'AGI' or 'superintelligence' race. In a TechCrunch interview, he said they never used the 'AGI' term and 'superintelligence' has become outdated. 'Next time we’ll switch to something else.' He criticized the lack of definition: 'What is superintelligence? I don’t know. It’s not a very useful word.'

AMI Labs' 'World Model' Strategy

While AMI Labs has no product yet, it’s courting robotics, manufacturing, and electronics players. LeBrun explained world models predict and interact with the real world, unlike LLMs that predict text. 'If you nudge a glass off the table, you already know it will spill—world models capture this intuition.'

Robotics and the Lack of Real-World Understanding

LeBrun sees a major opportunity in robotics. Current robots run fixed routines, he said. 'Hardware is advanced, but there’s no brain.' He highlighted the need for context-aware AI to prevent robots from harming children. In healthcare, he compared today’s AI to a doctor trained only on textbooks, noting LLMs cover only 1% of the field.

Asia-Focused Strategy

LeBrun is targeting South Korea for its advanced robotics, semiconductors, and manufacturing industries—areas the first wave of AI barely touched. He praised Korea’s early internet adoption and its aggressive AI investment. 'Korea should invest in physical AI too,' he said. SBVA CEO JP Lee emphasized Korea’s 'global leadership in AI and chip industries.'

AMI Labs' world model approach could redefine robotics and physical AI. Korea’s rapid adoption skills support this strategy, but the lack of a product means long-term market impact remains uncertain.