Meta’s “superintelligence” AI team appears to be largely made up of OpenAI, and has their work cut out for them
Meta’s path to AI domination seems to be paved with other AI companies and their talent.
Meta’s plan to rival OpenAI seems to be largely dependent on OpenAI itself.
The social media company has been assembling a “superintelligence” team to boost its AI chops, help it reach AGI, and allow it to keep dominating advertising. To furnish that team, this month Meta has tried to poach more than 45 AI researchers from OpenAI alone, The New York Times reports, offering starting packages as high as $100 million apiece. Already the social media behemoth has hired at least four OpenAI researchers, including Trapit Bansal, as well as Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai.
“At least, so far, none of our best people have decided to take them up on that,” OpenAI CEO Sam Altman had said earlier this month.
The Verge reports that at an internal all-hands meeting held this week at Meta, its employees had questions about these reported $100 million offers that Altman had described on a podcast, which executives tried to play down:
“Sam is just being dishonest here,” Andrew Bosworth, Meta’s CTO, said at the meeting when asked about Altman’s remarks. “He’s suggesting that we’re doing this for every single person… Look, you guys, the market’s hot. It’s not that hot.”
Meta leadership has even floated “de-investing” from Meta’s Llama and instead embracing AI models from competitors including OpenAI, according to NYT, though they haven’t made any final decisions yet.
Meta, of course, has been using its deep pockets to get non-OpenAI talent. That includes its $14 billion investment in Scale AI and its founder, Alexandr Wang, and hiring people from companies like Google, Sesame, and Safe Superintelligence.
Meta is also talking with AI voice startup PlayAI for a potential company and talent acquisition.
Low-engagement Llama?
Meta has boasted publicly about having 1 billion Meta AI users — the magic number it waits to hit before monetizing new products. The company has been spreading its AI all over its products, and only released a stand-alone “Meta AI” app a few months ago. But new details emerging from Meta executives raise some red flags about how many people are actually using it.
According to a report from The Verge on this week’s all-hands meeting at Meta, it may be less than they are saying publicly:
“Bosworth wasn’t the only Meta exec to mention OpenAI during the internal meeting. CPO Chris Cox also acknowledged that, while Meta AI has one billion monthly users, engagement ‘is not nearly as deep as the way that people are using ChatGPT.’ The standalone Meta AI app has only 450,000 daily users, he told employees, and ‘a lot of those folks’ are using it to manage their Ray-Ban Meta glasses.”
Cox also reportedly said at the meeting that they weren’t chasing AI-powered productivity tools like Google, Anthropic, and OpenAI, but would focus on “entertainment, on connection with friends, on how people live their lives, on all of the things that we uniquely do well.”
