Meta’s AI development has hit a roadblock, prompting CEO Mark Zuckerberg to take drastic measures. After a year of technical setbacks with their ‘state-of-the-art’ Llama 4, Zuckerberg is forming a secretive ‘superintelligence group’ comprising 50 handpicked AI researchers. This initiative comes with a hefty $15 billion investment for a 49% stake in Scale.ai, a controversial data labeling startup.
Scale.ai has faced accusations of systemic wage theft and exploiting third-world laborers for AI training data. The company is also contracted to support the Pentagon’s weapons automation program. Meta’s own data practices have come under fire, including scraping over seven million copyrighted books to train Llama, resulting in a major class action lawsuit.
Zuckerberg’s goal is to develop the first AI system surpassing human brain capabilities, a feat many researchers consider ‘very unlikely.’ The investment in Scale.ai is Meta’s largest external investment to date. Scale.ai’s co-founder, Alexandr Wang, will join Zuckerberg’s elite team, bringing his connections to the US defense industry.
Wang, known as the ‘world’s youngest self-made billionaire,’ has been vocal about an ‘AI war’ with China and has secured lucrative deals with the Pentagon. His involvement aligns with Zuckerberg’s history of collaboration with the US military and his previous stance on data privacy.
The ‘superintelligence’ initiative has drawn comparisons to the Marshall Plan, with tech critic Ed Zitron noting that Meta is running out of training data and resorting to Scale.ai’s labor practices. The investment underscores Zuckerberg’s aggressive push into AI, despite previous setbacks like the metaverse project.
Meta’s financial muscle, including its ad-based revenue stream and plans for a ‘multi-gigawatt data center,’ is being used to lure top talent to the secretive project. The company’s ability to offer seven- to nine-figure sums to recruits highlights the scale of Zuckerberg’s ambition in the AI race.
