Mark Zuckerberg Shares A Confusing Vision For Ai 'superintelligence'

Mark Zuckerberg has spent nan past respective months and several cardinal dollars recruiting salient AI researchers and executives for a caller "superintelligence" squad astatine Meta. Now, nan Meta CEO has published a lengthy memo that attempts to laic retired his large scheme for utilizing nan company's immense resources to create "personal superintelligence."

In the memo, which sounds much for illustration a manifesto than a strategical business plan, Zuckerberg explains that he's "extremely optimistic that superintelligence will thief humanity accelerate our gait of progress." The technology, according to him, "has nan imaginable to statesman a caller era of individual empowerment wherever group will person greater agency to amended nan world successful nan directions they choose."

Zuckerberg, who has antecedently expressed a desire to build artificial wide intelligence, ne'er defines "superintelligence." Nor does nan 616-word memo explicate really Meta plans to create specified a technology, what it mightiness thief group execute aliases why anyone should spot nan institution to build it. Instead, he implies that Meta will beryllium a amended steward of this non-specifically powerful AI than "others successful nan industry" who expect "humanity will unrecorded connected a dole of its output."

Left unsaid by Zuckerberg, is nan truth that nan memo comes astatine a clip erstwhile he's been quickly reorganizing Meta's AI teams. Last month, nan institution invested $14.8 cardinal into Scale AI, a move that allowed it to bring Scale CEO and laminitis Alexandr Wang into nan company. The 28-year-old laminitis is now Meta's Chief AI Officer successful complaint of its superintelligence efforts.

Meta has besides been connected a hiring spree for nan effort, and has reportedly been offering salient researchers eight- and nine-figure salary packages to travel to Meta. In caller weeks, nan institution has successfully recruited high-profile talent from Apple and OpenAI, including Shengjia Zhao, who helped created GPT-4. Zhao announced past week that he will return connected nan domiciled of "chief intelligence of Meta superintelligence labs." Just yesterday, Wired reported that Meta has precocious turned its recruiting efforts to Thinking Machines Lab, an AI startup founded by erstwhile OpenAI CTO Mira Murati, and that successful astatine slightest 1 lawsuit it made an connection worthy much than $1 cardinal complete respective years. (Meta PR said immoderate specifications of that study were "off.") All that is connected apical of nan $72 billion Zuckerberg has said Meta plans to walk connected AI infrastructure.

Driving each this is that Zuckerberg has reportedly grown progressively disappointment by Meta's ain generative AI efforts. The institution has had to delay its larger "Behemoth" Llama 4 exemplary by months. Llama's struggles person besides reportedly caused Zuckerberg to mobility whether Meta's AI efforts should stay unfastened source, according to CNBC.

It's besides apt nary coincidence Zuckerberg's rambling manifesto comes hours earlier nan institution is scheduled to study net and show analysts much astir its plans to walk billions of dollars connected caller AI efforts.

Meta's CEO besides intelligibly sees AI power arsenic an opportunity to extremity nan company's reliance connected mobile platforms, particularly Apple, which he believes person been capable to exert excessively overmuch power via their app stores. In his memo, he explains that "personal devices for illustration glasses … will go our superior computing devices." A early wherever smart glasses are much important than smartphones would, of course, beryllium highly convenient for Meta, which has spent nan past respective years building smart glasses.