source: techcrunch ai: why ami labs’ alexandre lebrun won’t call his ai ‘agi’ or ‘superintelligence’
level: business
alexandre lebrun, ceo of yann lecun's world model startup ami labs, refuses to use terms like agi or superintelligence. he says the industry has shifted from agi to superintelligence without clear definitions. lebrun finds the words unhelpful and expects the next buzzword to replace them soon. his stance comes as ami labs, still pre-product, scouts partners in robotics, manufacturing, and electronics during a visit to seoul for icml.
lebrun explains that world models predict the next state of the physical world, unlike large language models that predict text. he sees them as complementary, with llms handling language and world models providing real-world context. current robots run fixed routines and lack awareness, making them unsafe outside controlled settings. even basic context awareness would be a major advance, he says, pointing to an incident where a robot kicked a child at a public event.
ami labs needs real-world environments to train its models, pushing lebrun toward asia for its hardware industries and fast adoption. south korea's robotics, semiconductor, and manufacturing sectors, plus government ai funding, attract him. local investor jp lee notes korea's quick adoption of new tools and urges continued investment in physical ai. ami labs raised over a billion dollars but has no product or timeline, with lebrun promising a surprise when ready.
why it matters: rejecting hype labels helps focus on practical physical ai that could make robots safer and more context-aware in real-world settings.
source: techcrunch ai: why ami labs’ alexandre lebrun won’t call his ai ‘agi’ or ‘superintelligence’