Driving the surge in interest in generative AI is the conviction that the tech industry is progressing towards achieving superhuman, god-like intelligence.
OpenAI’s objective is explicitly to create artificial general intelligence (AGI), shared by other industry leaders such as Demis Hassabis from Google’s AI efforts. Joining this pursuit is Meta CEO Mark Zuckerberg, who, while lacking a defined timeline or precise definition for AGI, is determined to build it. Simultaneously, he is reshaping Meta’s AI research group, FAIR, integrating it with the team developing generative AI products across Meta’s applications, aiming for more direct application of AI breakthroughs to the company’s vast user base.
“We’ve come to this view that, in order to build the products that we want to build, we need to build for general intelligence,” Zuckerberg explains in an exclusive interview. He emphasizes the importance of conveying this vision, acknowledging that the most skilled researchers are drawn to ambitious challenges.
Zuckerberg openly addresses the fierce competition for AI talent, a competition intensified by the scarcity of experts in the field. Companies vie for a limited pool of researchers and engineers, with compensation packages reaching over $1 million annually. CEOs, including Zuckerberg, actively engage in recruitment efforts and retention strategies.
Beyond talent, computing power for training and running large AI models is another critical resource. Zuckerberg reveals Meta’s readiness to deploy significant resources, announcing plans to own more than 340,000 Nvidia H100 GPUs—the industry’s preferred chip for generative AI—by the end of the year. He highlights Meta’s unprecedented scale in this regard, potentially exceeding any other individual company.
Reflecting on AGI, Zuckerberg admits the lack of a clear definition or timeframe for its realization. He envisions AGI’s arrival as a gradual process, emphasizing the importance of intelligence breadth—reasoning and intuition. Meta’s expanded focus on AGI is influenced by the release of Llama 2, its latest large language model, and the company’s belief in coding’s structural importance for AI models’ understanding and logic.
The debate over open versus closed control of AGI is a contentious topic, as seen in the recent struggles of OpenAI. Zuckerberg, wielding considerable power at Meta, leans towards an open-source approach. He sees challenges in concentration if valuable technologies are closed, advocating for openness to address issues related to unequal access and opportunity.
Despite acknowledging safety concerns, Zuckerberg emphasizes the strategic alignment of the biggest AI companies, often calling for regulations while protecting their leads. The question of whether Meta’s AGI will be open source remains undecided, contingent on safety and responsibility considerations.
In the broader context of Meta, Zuckerberg’s AGI initiative appears somewhat at odds with the company’s recent focus on the metaverse. He refutes claims of a pivot, affirming continued commitment to Reality Labs and the metaverse, with substantial annual spending. Generative AI is expected to play a pivotal role in Meta’s hardware endeavors.
Zuckerberg envisions a future where AI-generated virtual worlds, populated by AI characters, coexist with real individuals. A new platform is slated to launch this year, allowing users to create and distribute their AI characters across Meta’s social apps, potentially posting content to Facebook, Instagram, and Threads.
While Meta remains a metaverse giant and the world’s largest social media company, it is now venturing into AGI development. Zuckerberg’s narrative centers around the overarching mission of “building the future of connection,” signaling a future where human-AI interactions become increasingly prevalent, whether society is fully prepared for this transformative shift or not.
