Mark Zuckerberg’s Shift on Open Source AI

An outdoor photo showing someone in a navy blue henley t-shirt giving a presentation or speech. They are gesturing with one hand raised while speaking, with a blurred natural background of trees and foliage behind them. The lighting appears to be natural daylight.

Mark Zuckerberg’s Vision for AI Superintelligence

Mark Zuckerberg shared a memo outlining his vision to build AI “superintelligence.” In the memo, he hinted that the pursuit of more powerful AI might require the company to be more selective in what it open sources, citing “safety concerns.” This statement stood out, given Zuckerberg’s previous emphasis on open source AI as central to Meta’s strategy.

During Meta’s earnings call, Zuckerberg acknowledged a potential shift in the company’s approach to open sourcing AI models. He downplayed the significance of the change, stating that Meta has always open sourced some models while keeping others proprietary. However, he noted that the company is wrestling with the practicality of sharing large models and the potential safety concerns associated with approaching real superintelligence.

Zuckerberg’s current stance differs from his previous memo, “Open Source AI is the Path Forward,” where he argued that open source AI is crucial for both Meta and developers. He believed that open sourcing models wouldn’t give away a massive advantage, as AI development is highly competitive. Additionally, he thought open source AI would be safer, as it promotes similar generations of models, allowing governments and institutions to check bad actors.

While Zuckerberg stated that Meta would continue to open source some of its work, his recent comments suggest that the company’s “superintelligence” could be less open in the future. This potential shift has sparked debate about Meta’s commitment to open source AI and its implications for the development of AI superintelligence.