Mark Zuckerberg said that AI agent progress has been slower than expected in an internal town hall

5 hours ago 1

Meta CEO Mark Zuckerberg

Meta CEO Mark Zuckerberg Louis Grasse/Reuters

Superintelligence at Meta is coming. But it's going to take a bit of time and elbow grease, first.

That's what the company's CEO Mark Zuckerberg told employees at an internal town hall on Thursday, according to two sources who listened in.

Meta is pouring vast resources into AI, but progress on AI agent technology hasn't moved as quickly as the company expected, Zuckerberg said, in comments first reported by Reuters.

Meta's latest internal reset shows the tension at the heart of the company's evolving AI strategy. Zuckerberg and his team are racing to build AI models and pouring tens of billions of dollars into talent and infrastructure. But the company is also learning that moving fast has costs.

As Meta pushes employees to accelerate AI development, it's increasingly having to balance speed with trust, morale, and the buy-in of the workforce expected to build it.

Even so, Meta still remains on a "journey to superintelligence," Zuckerberg told staff in the town hall, adding that the company expects to see some benefits within the next three to six months according to an employee on the call. That journey will take hard work, Zuckerberg said, pointing to just how competitive the AI landscape has become,

Meta declined to comment.

Meta is also walking back one of the more contentious pieces of its AI push. Chief technology officer Andrew Bosworth told the town hall that the company's AI training program, which used employees' keystrokes and mouse movements to train its models, will be opt-in only if and when it resumes, a person on the call said.

Business Insider previously reported that the mandatory version sparked a backlash from employees who were uncomfortable with their activity being recorded. Meta paused the program last month after an internal leak exposed employee conversations and keystrokes to their colleagues.

Bosworth acknowledged the rollout had damaged morale and trust within the company. But he said it had also generated more useful data than expected, enough, he suggested, that Meta hadn't needed to deploy it as broadly as it did.

The reversal echoes another recent retreat: last month, Meta gave engineers the option to leave its Applied AI task force, after having reassigned thousands of them to the unit — a climbdown some employees dubbed "the undraft."

That move came weeks after Bosworth warned staff that morale was "probably one of the worst it's ever been" in Meta's 20-year history — and after Meta laid off 10% of its staff, roughly 8,000 people, in May.

Have a tip? Contact Charles via email at [email protected] or on Signal and WhatsApp at 628-282-2811. Contact Pranav via email at [email protected] or on Signal at 408-905-9124. Use a personal email address, a nonwork WiFi network, and a nonwork device; here's our guide to sharing information securely.

Read next

Charles Rollet is BI's tech correspondent in San Francisco. Prior to joining BI, Charles worked at TechCrunch covering startups and VC. Charles is based in the Bay Area, where he enjoys hiking with his dogs. You can contact Charles securely on Signal at charlesrollet.12 or +1-628-282-2811.

Pranav Dixit is the Meta Correspondent at Business Insider based in the San Francisco Bay Area. He writes about Meta’s products, policies, and internal workings while examining how the company’s decisions shape how billions of people connect and communicate.Previously, Pranav was the India-based technology correspondent for BuzzFeed News, covering the impact of Silicon Valley’s largest companies on the culture, society, and politics of more than a billion people in South Asia. He has also been a senior news editor at Engadget and ran technology coverage at the Hindustan Times, one of India’s largest national newspapers.Pranav’s reporting has shed light on the human consequences of Big Tech’s quest for growth in emerging markets, and sparked widespread conversations about the impact of American technology companies on the Global South. In 2019, he won Syracuse University’s Mirror Award for a boots-on-the-ground feature about how WhatsApp misinformation sparked gruesome lynchings in rural India. He has also reported from Kashmir, a volatile geopolitical hotspot, documenting the world’s longest-running internet shutdown.His work has been widely cited by major national and international publications, and he has been featured on the BBC, Al Jazeera, and podcasts such as Vox Media’s Land of the Giants to discuss his work. He has also spoken in journalism classes including at UC Berkeley’s graduate journalism program. His writing has appeared in The Guardian, Vox, Time, The Information, and Al Jazeera.Pranav moved to the United States in 2021 from New Delhi, India, to be a fellow at Harvard University’s Nieman Foundation for Journalism, where he studied the evolution of the American tech press and ways newsrooms around the world can cover technology and society more effectively.Got a tip about Meta or anything else in Silicon Valley? Contact Pranav via encrypted messaging app Signal (+1408-905-9124), or email him at [email protected] or [email protected]. You can also reach him on WhatsApp at +857-753-3949 or DM him on X (@PranavDixit) or BlueSky (@pranavdixit.bsky.social).Pranav keeps sources anonymous. Please use a non-work device to reach out.Expertise: Meta, Facebook, WhatsApp, Llama, AI, Threads, Instagram, Mark Zuckerberg, social media, platforms, immigration

Read Entire Article
| Opini Rakyat Politico | | |