Mark Zuckerberg, the CEO of meta, has already stated that the company is currently working on artificial superintelligence (ASI). Focusing on the potential of such powerful systems, Zuckerberg acknowledged that it does not have to be fully-open sourced. Competitive and safety issues explain some of the main reasons why this could be one of the limitations to sharing.

Highlights:

• Zuckerberg confirms Meta is pursuing artificial superintelligence (ASI).
• Safety is a primary challenge for superintelligence development.
• Meta will invest heavily in computing power for superintelligence research.
• Open-sourcing may not apply to superintelligence-level systems.
• Safety and competition drive potential limits on sharing superintelligence.

Zuckerberg declared the development of superintelligence, or the supremacy of AI over human cognition, as one of the central and long-term goals of Meta. He recognized the existence of major technical challenges especially in development and deployment in a safe manner. The focus of the superintelligence strategy at Meta is to develop advanced computing infrastructure.

In the past, Meta has been an open-source AI advocate. But Zuckerberg made the statement that this might not be the case with superintelligence. He quoted the principal safety hazards at uncontrolled access. Another business factor is to have a competitive position. Meta pursues responsible development, yet this presupposes the fine management of the sophisticated superintelligence.

The statement by Zuckerberg makes it clear that Meta will focus on two things, both discovering the potential of super intelligence and addressing its major safety concerns. Whether it will be open or under inner control will be a crucial issue concerning superintelligence research.