AI godfather Yann LeCun leaves Meta to start his own company

Just a few weeks ago, one of the ‘godfathers’ of artificial intelligence was presented with an award by King Charles at St James’s Palace for his work in the field of artificial intelligence (AI).

Professor Yann LeCun was recognized along with six other laureates for his contributions to the field, recognized as advances in deep learning.

But LeCun is at odds with some in the AI ​​industry over the future of this generation-defining technology.

Now, he has announced that he is leaving his position as Meta’s chief AI scientist to start a new company, fully focused on the idea of ​​”advanced machine intelligence.”

During his 12 years at the company, Professor LeCun won the prestigious Turing Award and witnessed several waves of excitement around AI. In particular, we’ve seen a recent boom in generative AI that accelerated with rival OpenAI’s announcement of ChatGPT in late 2022.

But his resignation comes amid speculation that the AI ​​boom could come to an abrupt end if the so-called “AI bubble” of soaring valuations and soaring spending bursts.

Investors, analysts and even big tech executives like Google CEO Sundar Pichai say the market correction on the AI ​​sector will ripple through the economy.

Professor LeCun announced his plans to step down from Meta on Wednesday, after more than a week of rumors and reports about his departure.

In a series of posts on Threads, he thanked the company’s founder Mark Zuckerberg and highlighted the company’s Fundamental AI Research (FAIR) lab as its “proudest non-technical accomplishment.”

“As many of you have heard through rumors and recent media articles, I am leaving Meta after 12 years, including 5 years as Founding Director of FAIR and 7 years as Chief AI Scientist,” he wrote.

“The impact FAIR has had on the company, the AI ​​field, the technical community, and the wider world has been remarkable.”

For many years, the institute has focused on developing systems and technologies that advance machine learning and translation.

But like much of the industry, Meta is looking to focus much of its research and spending on large-scale language models (LLMs), the core systems for generative AI tools like chatbots and image generators.

Professor LeCun suggests that LLM is not very useful when trying to create AI systems that rival human intelligence.

Instead, he wants to pursue what he calls “advanced machine intelligence.”

It primarily trains AI models using visual learning, trying to replicate the way children and baby animals learn.

This is different from LLM, where you are fed a huge amount of existing data and are asked to generate results based on that data and prompts.

Professor LeCun said he would continue to have a relationship with Meta even if he formed a new company, adding that Meta would become a partner in his new company if he added a post about his departure.

But reports suggest he is moving further and further away from the company’s desired approach.

Unlike AI godfathers Jeffrey Hinton and Joshua Bengio, Professor LeCun questions the idea that AI could pose an existential threat to humanity.

In 2023, he called such concerns “ridiculous.”

“Will AI conquer the world? No, this is a projection of human nature onto machines,” he told the BBC.

However, some question the characterization of Professor LeCun as an industry maverick or visionary.

“Jan LeCun has undoubtedly made a real contribution to AI, and I’m glad to see him speaking out again about the limitations of LLMs,” AI expert and professor Gary Marcus said in a recent blog.

“But he has also systematically dismissed and ignored the work of others for years.” Among them was his own, whose work Professor LeCun suggested was often overlooked.