Zuckerberg’s vision for AI involves “personal superintelligence”
Zuckerberg has made it clear that his approach to artificial intelligence differs from others in Silicon Valley. The work that Meta’s Superintelligence Labs and its newly recruited superintelligence team will focus on is developing AI to improve people’s personal lives – not to automate work, which has become a growing concern as more companies lay off employees due to AI adoption in the workplace.
“I am extremely optimistic that superintelligence will help humanity accelerate our pace of progress. But perhaps even more important is that superintelligence has the potential to begin a new era of personal empowerment, where people will have greater agency to improve the world in the directions they choose,” Zuckerberg said.
Currently, superintelligence does not exist. But if it ever becomes a reality, it would be an AI system capable of teaching itself, learning from mistakes, writing code as well as or better than humans, and potentially even making emotional or ethical judgments.
Superintelligence is fundamentally different from where AI stands today. The tools we currently interact with such as AI chatbots like ChatGPT are far from reaching that level of autonomy and capability.
The biggest question surrounding Meta’s ambitious AI project is: How will it differ from the AI we are experiencing today? Will it move us closer to artificial general intelligence (AGI), or will it land somewhere between current AI capabilities and true AGI?
It’s also unclear whether superintelligence will eventually power devices like smartphones and smart glasses, or if the next generation of AI models will evolve beyond current multi-modal systems which understand text, video, and images and focus instead on continual learning and deeper contextual understanding that extends beyond static datasets.
Imagine AI models capable of powering humanoids – machines that can adapt, reason, and interact like humans in the physical world. For now, however, none of the AI models from the top research labs have these capabilities.
‘Superintelligence’ AI models are likely to remain closed
As the CEO of a $1.75 trillion company, Zuckerberg did not outline specific products or applications Meta plans to launch, but he did hint at how the company will release its AI models as it pursues “superintelligence.”
“We believe the benefits of superintelligence should be shared with the world as broadly as possible. That said, superintelligence will raise novel safety concerns. We will need to be rigorous about mitigating these risks and careful about what we choose to open source.”
Imagine AI models capable of powering humanoids – machines that can adapt, reason, and interact like humans in the physical world (AI-generated image: Microsoft Designer)
Zuckerberg carefully chose his words and specifically emphasized “open source,” signaling a shift in Meta’s strategy. Like its competitors, Meta now appears to be moving toward keeping its AI models closed. While its existing Llama models are partially open source, they are not fully open. Closed models give the company greater control over how it develops and monetises its AI technology.
Although it remains unclear how Meta’s superintelligence AI will differ from existing tools and chatbots, the focus on “personal superintelligence” suggests the company may deliver the next chapter in AI through its own ecosystem, including apps like Facebook, Instagram, and WhatsApp, as well as hardware like smart glasses, augmented reality (AR) devices, and mixed-reality headsets.
Meta spends billions on top AI talent
Zuckerberg has been bullish on artificial intelligence and is committed to spending billions of dollars on building both AI infrastructure and a global team of top AI researchers. In fact, he personally spent months assembling a list of elite AI engineers and scientists, recruiting them to join Meta Superintelligence Labs, the unit dedicated to creating world-class AI models that could surpass human intelligence.
Meta has been racing to keep up with competitors like OpenAI, Google, and Anthropic.
To achieve its ambitions, Meta is aggressively poaching AI talent from rival companies, offering eye-popping compensation packages – in some cases as high as $200 million. The company is clearly trying to win the AI race with money.
Meta, which owns Facebook, Instagram, and WhatsApp, is leaving no stone unturned in its pursuit of AI dominance, investing billions in AI research and development. However, some industry observers question Meta’s current AI strategy. Its Llama models are not the best, and there have been delays in launching its flagship model, known internally as “Behemoth”, have raised questions on Meta’s AI development strategy. Meanwhile, OpenAI is expected to release GPT-5 as soon as August.
Earlier this month, Meta announced a $14.3 billion investment in Scale AI and hired its CEO, Alexandr Wang, to lead its Superintelligence team. The 28-year-old will now oversee a group composed of top-tier AI researchers, including alumni from PhD programs at UC Berkeley and Carnegie Mellon. Many researchers from Apple and OpenAI have already joined the team, including Ruoming Pang, who previously led Apple’s AI models division and is reportedly earning a compensation package in the hundreds of millions over several years.
Investors will watch for Meta’s new AI strategy to develop
Zuckerberg told analysts Wednesday during a second-quarter earnings that its newly setup Superintelligent AI team are “the absolute best and most elite talent-dense team” that can access the resources they need from a “leading compute fleet.” Meta made it clear that winning the AI war comes with a heavy price tag.
“When we take a technology, we’re good at driving that through all of our apps and our ad systems,” Zuckerberg said. “There’s no other company that is as good as us at kind of taking something and getting it in front of billions of people.”
The company said it expects its total expenses for 2025 to come in the range of $114 billion and $118 billion. Meta’s strong Q2 earnings surpassed Wall Street’s estimates, sending the stock soaring in after-hours trading. The company reported revenue of $47.52 billion, compared to expectations for $44.3 billion.
Investors, however, are not in a panic mode with Meta’s expensive endeavors. The company previously bet heavily on the metaverse, investing billions of dollars but the metaverse never truly took off.
Meta’s grand vision for AI dominance also includes building hardware designed around artificial intelligence. Its Ray-Ban smart glasses, AI-ready eyewear that allows users to make calls, take photos, and perform other functions, have performed well commercially.
“I think personal devices like glasses that can see what we see, hear what we hear, and interact with us throughout the day will become our main computing devices,” Zuckerberg said in a social media video posted on X.
Ray-Ban maker EssilorLuxottica reported in its most recent earnings report that sales of Ray-Ban Meta smart glasses more than tripled in the first half of the year.
Meta, which owns Facebook, Instagram, and WhatsApp, is leaving no stone unturned in its pursuit of AI dominance, investing billions in AI research and development(File photo) 
Imagine AI models capable of powering humanoids – machines that can adapt, reason, and interact like humans in the physical world (AI-generated image: Microsoft Designer)
Share
