|

Meta’s Scoping Out a Superintelligence Lab

As ChatGPT’s partial outage yesterday goes to show, AI companies are still struggling to scale up to artificial general intelligence.

Photo of Meta's Mark Zuckerberg
Photo via Andrej Sokolow/dpa/picture-alliance/Newscom

Sign up for smart news, insights, and analysis on the biggest financial stories of the day.

It’s your first day at Meta, and your desk is directly next to Mark Zuckerberg’s. That’s what’s happening for new AI employees at Meta, where Bloomberg reports Zuck is personally assembling a squad to lead a new lab that’ll build AI that’s smarter than humans, aka “superintelligence.” 

Meta’s bringing on Scale AI CEO Alexandr Wang to lead the lab, reports The New York Times, as the tech giant prepares to pour $10 billion into the AI startup. ScaleAI’s business model involves helping other companies, including OpenAI and Microsoft, build AI. 

Machine (Still) Learning

If Meta can crack superintelligence, it’d set the tech company apart from rivals that are stuck trying to build “artificial general intelligence” (AGI) that can match, but not go beyond, the human brain’s capabilities. 

As ChatGPT’s partial outage yesterday goes to show, AI companies are still struggling to scale up to AGI. During the blackout, OpenAI’s chatbot experienced more errors and delays than usual — like an AI brain freeze. The chatbot also struggled to keep up with queries in March, when Sam Altman said its “GPUs are melting” after a flood of users asked it to make Studio Ghibli-style selfies.

ChatGPT accelerated the AI race when it came on the scene, but its struggles to scale up could create a lucrative opening for tech rivals willing to throw their multi-billion-dollar hats in the ring:

  • A $10 billion investment into ScaleAI would be Meta’s biggest external investment in artificial intelligence so far as it earmarks tens of billions toward developing the technology this year. It previously focused on internal efforts, like its Llama AI model and its Meta AI app.
  • Meta missed the chance to buy AI company DeepMind, which Google scooped up for $600 million in 2014 and has made integral to its AI strategy. Meanwhile, Amazon has poured $8 billion into Anthropic, and Microsoft has tied its cart to OpenAI, investing $13 billion in the startup so far. Apple has taken a smaller role in the AI race.

Talent Competition: Some of Big Tech’s billions spent on AI are going toward hiring, as securing the best talent becomes a competitive sport. Not only is Zuck said to be personally involved in AI recruiting at Meta, but Meta has been sending out seven- to nine-figure offers to lure AI researchers away from OpenAI and Google. Meta is reportedly still losing talent to competitors that are also spending billions on recruitment. Last year, Google paid $3 billion on hiring specialists and execs and licensing tech from Character.AI. Winning the AI race could hinge on HR.

Sign Up for The Daily Upside to Unlock This Article
Sharp news & analysis on finance, economics, and investing.