Meta is creating a new AI lab to pursue ‘superintelligence’ | Technology News

The new lab is part of a larger reorganization of Meta’s AI efforts, the people said. The company, which owns Facebook, Instagram and WhatsApp, has recently grappled with internal management struggles over the technology, as well as employee churn and several product releases that fell flat, two of the people said.

Mark Zuckerberg, Meta’s CEO, has invested billions of dollars into turning his company into an AI powerhouse. Since OpenAI released the ChatGPT chatbot in 2022, the tech industry has raced to build increasingly powerful AI. Zuckerberg has pushed his company to incorporate AI across its products, including in its smart glasses and a recently released app, Meta AI.

Staying in the race is crucial for Meta, Google, Amazon and Microsoft, with the technology likely to be the future for the industry. The giants have pumped money into startups and their own AI labs. Microsoft has invested more than $13 billion in OpenAI, while Amazon has plowed $8 billion into AI startup Anthropic.

The behemoths have also spent billions to hire employees from high-profile startups and license their technology. Last year, Google agreed to pay $3 billion to license technology and hire technologists and executives from Character.AI, a startup that builds chatbots for personal conversations.

In February, Zuckerberg, 41, called AI “potentially one of the most important innovations in history.” He added, “This year is going to set the course for the future.”

Meta and Scale AI declined to comment. Bloomberg earlier reported that Wang was joining the new Meta lab.

Superintelligence is regarded by leading researchers to be a futuristic goal of AI development. OpenAI, Google and others have said their immediate aim is to build “artificial general intelligence,” or AGI, shorthand for a machine that can do anything the human brain can do, which is an ambition with no clear path to success. Superintelligence, if it can be developed, would go beyond AGI in its power.

Meta has invested in AI for more than a decade. Zuckerberg created the company’s first dedicated AI lab in 2013, after losing out to Google in trying to acquire a seminal startup called DeepMind. DeepMind is now the core of Google’s AI efforts.

Since then, Meta’s research efforts have been overseen by its chief AI scientist, Yann LeCun, who is also a New York University professor. LeCun is a pioneer of neural networks, the technology that drives ChatGPT and similar systems.

After ChatGPT caused an explosion of interest in AI, Meta deployed additional resources to pursue the technology. It created a generative AI group, led by Ahmad Al-Dahle, a company vice president. LeCun’s research group also began working on what he saw as the next generation of AI.

LeCun, who was among three AI researchers who won the 2018 Turing Award, often called the Nobel Prize of computing, is highly respected across the field. But his views on AI differ from others in Silicon Valley. Whereas some believe that current technologies will reach AGI over the next few years, LeCun has said entirely new ideas are needed to reach this lofty goal.

One of Meta’s strategies for gaining ground in AI has been to “open source” its software, essentially giving away its AI code freely so that developers and others adopt its tools. The company released an open-source AI model, Llama, and its chatbot product, Meta AI. Meta AI was incorporated across Facebook, Instagram and WhatsApp, as well as in its Ray-Ban smart glasses. In May, Zuckerberg said more than 1 billion people used Meta AI every month.

More recently, Meta’s AI division has lost employees to rival companies, according to two people familiar with the matter. The departures were the result of a grueling pace of product development, infighting among team leaders and a tight labor market.

In April, Zuckerberg announced two new versions of Meta’s Llama AI models, which he claimed performed as well as or better than comparable models from OpenAI and Google, according to testing benchmarks compiled by Meta.

Soon after, outside researchers found that Meta’s benchmarks were designed to make one product look more sophisticated than it was. Some developers were incensed at what they saw as Meta’s trickery.

But not as incensed as Zuckerberg, who was upset that people thought he was trying to paper over the poor performance of the latest release, two of the people said.

Meta is now betting that Wang will help it get back into pole position in the AI race.

It must do so carefully. The Federal Trade Commission recently took Meta to trial in federal court over its acquisitions of Instagram and WhatsApp. An unusually structured investment deal with Scale AI could help Meta sidestep some of those concerns.

Wang founded Scale AI in 2016 alongside Lucy Guo, an engineer who was later fired by the company. Scale AI helped other businesses build AI technologies. It hired armies of contract workers to sift through vast amounts of data, labeling and “cleaning” the information so it could be used to train complex AI systems. Scale AI’s customers included OpenAI, Microsoft and Cohere, an AI startup in Toronto.

More recently, Scale AI has worked to build its enterprise and public sector businesses, dispatching consultants and engineers to work with companies and governments to help build software that uses AI.

Wang once lived in the same house as OpenAI CEO Sam Altman. In January, the two were photographed side-by-side at the Capitol for the inauguration of President Donald Trump. The next day, Scale AI placed an ad in The Washington Post, in which Wang called on Trump to increase investment in AI or risk falling behind China.

“Dear President Trump,” the ad said, “America must win the AI war.”