Meta Plucks Top Experts From OpenAI, Aims for 'Artificial Superintelligence'

On the same topic:

This week Meta concluded a blitz that in a matter of days rewrote the artificial intelligence sector, snatching eleven researchers from OpenAI, including some of the main architects of ChatGpt and the recent Gpt-4o model. These are figures that Mark Zuckerberg has personally courted for months, offering million-dollar contracts. According to the Wall Street Journal, some of them have signed for over $300 million in three years, confirming how hot the market for super talent in the sector is – and increasingly similar to the summer football transfer market.

The operation came a few days after another large investment by Meta, which has invested over 14 billion dollars in Scale AI, a company specializing in training artificial intelligence models . Its CEO, Alexander Wang, will lead the new Meta Superintelligence Lab, a research center created in less than a month with an explicit goal: to create a “superintelligence.” A strong word with an unclear meaning, behind which lies all the ambition of Meta, determined not to be left behind in the challenge with OpenAI, Google and Microsoft.

At the center of this cold war between big tech is a simple list, simply called “The List,” which includes a few dozen researchers specialized in machine learning, language model training, optimization techniques . They are in their twenties or thirties, they have obtained a PhD from Berkeley or Carnegie Mellon, they often know each other and have worked for the same companies, OpenAI and Google among them. We can see them as members of an intellectual elite in which the richest companies in the world are investing billions of dollars.

Sam Altman, CEO of OpenAI, followed Zuckerberg’s long courtship with irony and concern . First he assured that “none of our best men” had accepted his flattery (and millions of dollars); then, in the last few days, he assumed a serious tone: “Missionaries are better than mercenaries,” he recalled. As if it were strange that professionals could accept more advantageous proposals from competing companies.

While these top researchers are snatching contracts that are no match for the one just signed by Cristiano Ronaldo with Al Nassr, the crisis is hovering in the so-called “middle class” of programmers. In particular, the junior ones, who, as soon as they graduated, used to propose themselves to tech giants to start a promising career.

Microsoft just announced it would cut another 9,000 positions, and Google has made a similar move, as have many other tech companies in recent years. There are several reasons for the ongoing cuts, such as excessive hiring in 2020 and 2021, in the midst of the pandemic, but AI also has a bearing.

The CEO of Klarna, for example, a service for paying in installments for online purchases, recently said that he stopped hiring staff “a year ago,” precisely because AI is more than enough for many tasks. Again, the rhetoric is epic, Silicon Valley: many have noted that Klarna, in fact, continues to hire and has many open positions on its site. But AI optimizes and probably reduces the number of people needed in a company.

This problem especially affects the younger ones and entry-level positions, especially among programmers. In this sector, in fact, there are specific services (such as Cursor) capable of generating excellent quality code, which is then reviewed and integrated by a senior (and human) programmer.

So what? A crucial phase opens for Meta: after having assembled an elite team and allocated billions, he must demonstrate that he is able to transform all this into something concrete. And above all, he must explain to us what he really means by “superintelligence” .

More on these topics: