Mehta's chief AI scientist reportedly planning to leave the company to create his own startup,

according to the financial times, as the social media giant undergoes an AI overhaul.

For today's tech check, Deirdre Boso, looking into what the exit might say about Mehta's position

in the race, Deirdrejo, I'm so glad you're taking this up because I feel like it's a really big live

debate since earnings.

It is, and there's a lot of implications. So first of all, Yan Lukut, most people know this, but he's not

just another researcher. He is a pioneer of modern AI and one of its most outspoken skeptics. So he has

long dismissed the hype around generative AI and the race to super intelligence, and that

has increasingly put him at odds with Zuckerberg's work to commercialize Meta's AI. It has also put

him in a backseat to new flashy product hires like Alexander Wong. Now, it also signals just how

far meta has drifted from its early promise as a research powerhouse and its early wins like Lama, which when it was launched, it was one of the most capable open source models on the market.

Now, though, META is chasing the pack with product launches, expensive hiring sprees.

Lama, it has fallen way down in the LLM rankings, replaced by Chinese open source models.

And the result of this, investor doubt, that is creeping in.

Meta shares over the past three months, take a look.

They have massively underperformed the other mega caps.

What is next for Le Cun, now that could hint at a shift in the field itself.

According to the FT, his next venture will focus on something called

world models. That's AI that learns by observing the physical world. It is a slower burn

approach, more about understanding how intelligence actually works. And it's a clear contrast to the

bigger is better philosophy that is driving meta, open AI, and others, along with billions of dollars in CAPEX. Now, like other top AI names that have left big companies to build their own ventures, there's Miramir Maradi and Ilya Sutskever from Open AI. Lecun, he will likely have no problem raising money. We've seen billion dollar seed rounds from the others. Open question, though, guys, as to whether a commercial product will follow. We know it's not as easy for these research heads to monetize product.

So I was talking about this meta question with Brad Gersner last week. In Pebble, there was a big

...