SAN FRANCISCO — Mark Zuckerberg has spent the last nine months against the ropes as his company has made big cuts to its workforce and struggled to gain mainstream traction with its ambitious plans for virtual reality.
On Thursday, he told Meta employees how he planned to get the company back on track. In an all-hands meeting, Zuckerberg offered an explanation for recent layoffs and for the first time laid out a vision for how Meta's work in artificial intelligence would blend with its plans for the virtual reality it calls the metaverse.
Zuckerberg's talk was an attempt to rally staff after the most tumultuous period in his company's 19-year history. The chief executive said he made "tough decisions" about layoffs with the goal of "building a better technology company" that shipped better products, faster — something he believed Meta wasn't doing well as it swelled to more than 80,000 employees at the peak of the pandemic.
"I want us to use this period that's going to be a bit more stable in order to evolve and rebuild our culture," he said, according to two people who attended the meeting and shared remarks and a recording with The New York Times.
Zuckerberg delivered the remarks in a roughly half-hour address to thousands of employees at Meta's Menlo Park, California, campus. The talk, made on an outdoor pavilion the company calls Hacker Square, was also livestreamed to tens of thousands of employees around the world.
It was one of Meta's few major all-hands meetings over the last three years to be conducted in person, and included presentations from other Meta executives, including Andrew Bosworth, the chief technology officer, and Chris Cox, the chief product officer.
During the presentation, Cox detailed Meta's plans for making improvements to Reels, Instagram's short-form video product, to better take on TikTok, one of Meta's most formidable competitors.
Executives also spoke about Project 92, a long-rumored social app in development at Meta that will function similar to Twitter. The app, executives said, will work with other apps like Mastodon and Bluesky.
While Meta has aggressively worked on AI for several years, it has been slower than competitors like Google and Microsoft to turn that research into consumer products. Zuckerberg on Thursday detailed plans for artificially intelligent assistants that aid people across all Meta's apps, including WhatsApp, Messenger and Instagram.
He said Meta would work on creating artificial intelligence models that were accessible to more people than those of his company's competitors and, ultimately, would fit into his plans for the metaverse.
"Democratizing access to this has a bunch of value," Zuckerberg said, according to the two people who shared remarks with the Times. "But it's also aligned with the product vision of enabling a lot of different AIs instead of just trying to consolidate this ourselves into one singular AI that's going to try to rule everything."
He envisioned AI assistants that help people "create content to express yourself and your ideas so much better," or perhaps some artificially intelligent version of "a coach that gives you advice, encourages you."
AI agents could serve customers in products like WhatsApp, the globally popular messaging app that Meta has been focused on turning into an important tool for business owners and customer service. And every business could use a personalized AI algorithm.
"Different people have different interests, and we'll need a diverse array of AIs to represent all of these different interests," Zuckerberg said in the meeting.
To do that, the company is betting heavily on open source technology, which means it will share its work on artificial intelligence with researchers who want to build their own algorithms with what Meta has already done. The company has spent billions over the past decade building systems to run AI and attracting top researchers to work on some of the world's most difficult computer science questions around AI.
Meta has been criticized for its approach. Researchers and politicians outside the company say opening AI algorithms to many others could spawn malicious, automated and intelligent systems that accelerate the spread of misinformation. Those sophisticated algorithms, critics say, need to be tightly controlled.
In his address, Zuckerberg defended Meta's strategy. He said open-source software enables greater outside scrutiny of the technology because it can be seen by millions of technologists. He said working closely with outsiders' advances would make Meta's platforms better.
Zuckerberg also said he hoped for a world where people could build as many different AI programs as they wanted, rather than relying on a few provided by two or three large technology companies.
That does not mean Meta is backing away from its namesake metaverse plans, Zuckerberg said. Programs using new generative AI technology, he said, could eventually help people build new virtual world items and experiences. And he said the company planned to bring its AI assistant into the next version of its smart glasses. (Meta released a pair of smart Ray-Ban glasses in 2021, though sales have been sluggish.)
He also took a swipe at Apple's recently announced Vision Pro headset, $3,500 high-tech goggles that promised to usher in a new era of "spatial computing."
"I was really curious to see what they'd ship, and it's a good sign for our own development that they don't have any magical solutions to the laws of physics that we haven't already explored," he said in his remarks. Zuckerberg criticized the high-end materials and cost of the device, while noting that Meta had spent years bringing down the price of its headsets to an upcoming version that will start at $500.
"Their announcement really shows how our vision and values are different and what's at stake in shaping this platform," Zuckerberg said. "Our vision of the metaverse and presence is fundamentally social and about people interacting and feeling closer in new amazing ways. By contrast, every demo Apple showed was someone sitting on a couch by themselves."