Yann LeCun Calls Alexandr Wang ‘Inexperienced,’ Predicts Meta Exits

AI pioneer Yann LeCun is not offered on Mark Zuckerberg’s $14 billion guess on Alexandr Wang, the 28-year-old Scale AI cofounder recruited to guide Meta’s Super Intelligence Lab.

In a brand new interview with the Monetary Instances, Lecun, who was Meta’s chief AI scientist earlier than saying in November that he was leaving to kind his personal startup, mentioned Wang was “inexperienced” and did not totally perceive AI researchers.

“He learns quick, he is aware of what he would not know . . . There isn’t any expertise with analysis or the way you observe analysis, the way you do it. Or what can be enticing or repulsive to a researcher,” LeCun mentioned.

Wang was the crown jewel of Zuckerberg’s aggressive strikes within the AI talent war. Meta made a $14 billion investment in Scale AI, which included hiring Wang from the buzzy startup.

LeCun mentioned that Zuckerberg grew pissed off after disappointing progress on Llama, the corporate’s flagship, open-sourced AI mannequin.

Within the interview, LeCun mentioned that the AI staff “fudged” a number of the outcomes of Llama 4. On the time, Meta was criticized for doubtlessly gaming the outcomes of benchmark assessments. LeCun mentioned the episode soured Zuckerberg on Meta’s current AI staff.

“Mark was actually upset and principally misplaced confidence in everybody who was concerned on this,” he advised FT. “And so principally sidelined your entire GenAI organisation.”

As for his relationship with Wang, LeCun mentioned that despite the fact that the 28-year-old was briefly his boss after Zuckerberg’s AI reorg, he wasn’t actually directing him.

“You do not inform a researcher what to do,” LeCun advised the publication. “You definitely do not inform a researcher like me what to do.”

Meta’s new AI staff is ‘fully LLM-pilled’

LeCun mentioned Zuckerberg remained supportive of his views on the way forward for AI, however that the Meta CEO’s bigger hires are centered on massive language mannequin growth.

“I am certain there’s lots of people at Meta, together with maybe Alex, who would love me to not inform the world that LLMs principally are a lifeless finish on the subject of superintelligence,” LeCun mentioned. “However I am not gonna change my thoughts as a result of some dude thinks I am flawed. I am not flawed. My integrity as a scientist can not enable me to do that.”

LeCun has repeatedly argued that LLMs are too restricted and that to unlock the true energy of AI, a unique method is required. It is why his startup is reportedly known as Superior Machine Intelligence, the very method he has argued is healthier suited than LLMs.

He’ll function govt chair, not as CEO, of the brand new firm.

“I am a scientist, a visionary. I can encourage folks to work on attention-grabbing issues. I am fairly good at guessing what kind of expertise will work or not. However I am unable to be a CEO,” LeCun mentioned. “I am each too disorganised for this, and likewise too previous!”