"His startup is called Advanced Machine Intelligence (or AMI) and has hired Alex LeBrun, co-founder and CEO of medical transcription AI startup darling Nabla, as its CEO. Nabla disclosed LeBrun's new job in a press release and LeCun confirmed it in a brief post on LinkedIn. "Yes, AMI Labs is my new startup. I'm the Executive Chairman. And Alex LeBrun is transitioning from CEO of Nabla to CEO of AMI Labs!" LeCun wrote."
"AMI Labs is also reportedly seeking to raise €500 million (about $586 million) at a €3 billion valuation (about $3.5 billion) right out of the gate, before even launching, the Financial Times reported, citing people familiar with the dealmaking. Given the kind of money that VCs are throwing at AI startups founded by world-recognized AI scientists these days, that's not even an comparatively outrageous ask."
"The press release also confirms what everyone knew as well: that AMI Labs is working on world model AI. This is an alternative to LLMs where the AI attempts to understand its environment (aka the world) so it can simulate cause-and-effect and what-if scenarios to predict outcomes. World model creators believe it's the answer to LLMs' structural hallucination problems. LLMs can't be trusted never to fabricate info because it is their very nature to be "non-deterministic," aka creative."
Yann LeCun launched Advanced Machine Intelligence (AMI) and will serve as Executive Chairman while Alex LeBrun transitions from Nabla to become AMI's CEO. AMI Labs is seeking to raise €500 million at a €3 billion valuation as an early financing target. AMI focuses on world-model AI, which builds models of the environment to simulate cause-and-effect and predict outcomes. World models aim to reduce structural hallucinations by enabling what-if reasoning, offering an alternative approach to large language models. LeCun is a NYU professor, former Meta AI chief, and A.M. Turing Award winner, and other labs like Google DeepMind and World Labs pursue similar work.
Read at TechCrunch
Unable to calculate read time
Collection
[
|
...
]