‘AI Godfather’ Yann LeCun confirms establishing a new AI company

On LinkedIn on December 19, Yann LeCun confirmed that his new startup is called Advanced Machine Intelligence (AMI Labs). However, he will not run this new startup as CEO, instead hiring Alex LeBrun, co-founder and CEO of Nabla, a startup specializing in AI in the medical field, to take on the executive position.

“Yes, AMI Labs is my new startup. I am the executive chairman, and LeBrun will transition from CEO of Nabla to CEO of AMI Labs,” LeCun wrote.

In a subsequent press release, Nabla said LeBrun had left the company to join AMI Labs. Both companies will work together to develop the World Model – an ambition that LeCun has repeatedly mentioned with the ability to simulate the physical world, considered a different direction from the large language model (LLM) that companies currently pursue. LeBrun also confirmed the information on LinkedIn.

World Model is an alternative to LLM, in which AI tries to “understand” its environment (i.e. the world) and simulate cause-and-effect scenarios, providing hypothetical situations to predict outcomes. People like LeCun believe that this is the answer to the problem of “illusion” in the structure of the LLM, meaning that World Models will not fabricate information because their nature is “non-deterministic”, i.e. capable of creativity.

Previously, sources from Financial Times said AMI Labs is looking to raise about $586 million. This level helps the new company be valued at up to 3.5 billion USD upon launch, even before officially operating.

According to TechCrunch, considering the amount of money that venture capital funds are pouring into startups founded by leading AI scientists, this is not a very high number. For example, Thinking Machines Lab of former OpenAI CTO Mira Murati, was valued at $12 billion in its initial funding round last year; or Safe Superintelligence (SSI) of former OpenAI chief scientist Ilya Sutskever also raised billions of dollars after its founding.

LeCun, 65 years old, is a famous expert in the field of AI. He, along with professors Geoffrey Hinton and Yoshua Bengio, is known as the Godfather of AI. He joined Meta in 2013, heads FAIR – the team that created Meta’s first open source model called Llama – and is the company’s chief AI scientist. However, he left Mark Zuckerberg’s company last month to create his own startup specializing in “world simulation capabilities”.

Before leaving Meta, LeCun assessed that the current LLM is useful but does not have the potential to be as intelligent as humans. He believes that the field of artificial intelligence is being overhyped, especially the race to develop the LLM model. To enter the next revolution, companies need to “take a step back” to find what is missing in current approaches if they want to create human-level intelligent AI.