Behind the Scenes of Meta's 'Secret AGI Taskforce'
Mark Zuckerberg is reportedly building a team of AI specialists to achieve artificial general intelligence (AGI), recruiting from researchers and engineers.
As Meta's CEO, he has made recruiting for the confidential new team a priority, referred to internally as a superintelligence group, according to Bloomberg.
He believes Meta can surpass other tech companies in AGI – which represents the concept that machines can perform as effectively as humans across many tasks.
- AGI, or artificial general intelligence, is still just a hypothetical technology. The goal of AGI developers is to create an AI model that has the cognitive abilities of a human, including the human brain's ability to reason, learn, and solve a wide range of tasks. It is distinct from 'narrow AI', which specialises in solving specific tasks. Instead, developers what AGI to be extremely versatile.
Once Meta achieves that milestone, it could incorporate the capability into its range of products, including social media and messaging platforms, its Meta chatbot and AI-powered Ray-Ban glasses.
Meta now plans to hire approximately 50 people for the new team, including a new head of AI research, nearly all of whom he is recruiting directly.
According to reports, the creator of Facebook has already begun reorganising desks at the company's Menlo Park headquarters so the new staff will be seated near him.
Meta's plans with Scale AI
Mr. Zuckerberg is developing the team alongside a planned multi-billion dollar investment in Scale AI, a company that provides data services to help companies train their AI models.
Scale AI Founder Alexandr Wang is also anticipated to join the superintelligence group after a deal is finalised. The investment is expected to become Meta's largest external investment to date.
Zuckerberg has spoken publicly about making AI a priority for his company. In the past two months, he has embraced what insiders describe as "founder mode", characterized by an increasingly hands-on management approach, according to people familiar with his work.
His desire to personally manage the recruitment effort stems partly from frustration over the quality of and response to Llama 4, the newest version of Meta's large language model (LLM) that powers chatbots and other services.
The latest release in April proved disappointing to the CEO, who had consistently told Meta insiders he wanted the best AI offering by the end of the year.
His demands then created pressure on his AI-focused staff who were tasked with achieving those goals.
The model's performance has been questioned both internally by Meta's own leadership and externally by developers who viewed them as over-promising and under-delivering.
The impact of Meta's delayed "Behemoth" model
Meta subsequently delayed plans to release its largest model yet, known as 'Behemoth', which it had promoted as superior to competing models from OpenAI.
However, despite those claims, the leadership team grew concerned it did not sufficiently advance beyond previous models, according to reports.
This setback has reportedly prompted Zuckerberg to become more involved in Meta's AI expansion, which is beginning with the establishment of this new team.
According to Bloomberg, he created a WhatsApp group chat among senior leaders called 'Recruiting Party' to discuss potential targets for the team and has additionally compiled his own list of recruits and prefers to be the first point of contact during outreach.
Now, he hopes that with the new team, Meta will see enhancements to its Llama AI models.
It remains unclear how the Superintelligence Labs group will function alongside Meta's existing AI teams, but some employees are expected to transfer to the new unit, according to people familiar with the plans.
Meta's strengths and competition
Meta is competing against competitors like OpenAI and Google to become a market leader for AI, which already supports a significant portion of the company's advertising business.
The company has invested substantially in AI, allocating tens of billions of dollars in capital expenditures for projects this year and what Mark has assured would be "hundreds of billions" in years to come.
Bloomberg says that over lunches and dinners at his California residences in the past month, Mark has pitched AI researchers, infrastructure engineers and other entrepreneurs on joining Meta's team.
He has argued that, unlike competitors who are raising large funding rounds, Meta's advertising business is robust enough to finance the tens of billions of dollars needed to compete in the expanding AI space.
He also told potential recruits that Meta has sufficient cash flow to fund a multi-gigawatt data centre, which would give the company one of the most powerful server bases in the world.
Explore the latest edition of Technology Magazine and be part of the conversation at our global conference series, Tech & AI LIVE.
Discover all our upcoming events and secure your tickets today.
Technology Magazine is a BizClik brand
