Meta Considers Major Shift from Open-Source to Closed AI Models Under New Leadership
Meta explores a potential transition from open-source to closed AI models.
Key Points
- • Meta's lab is discussing shifting from open-source Behemoth to a closed model.
- • Alexandr Wang, the new chief AI officer, leads the discussions.
- • Meta has invested $14.3 billion in Scale AI, Wang's startup.
- • Any strategy change would need CEO Mark Zuckerberg's approval.
Meta's superintelligence lab is at a pivotal crossroads, deliberating a shift in its artificial intelligence strategy that could lead to abandoning its open-source model, Behemoth, in favor of a proprietary approach. This discussion is spearheaded by Alexandr Wang, Meta's newly appointed 28-year-old chief AI officer, who has previously co-founded Scale AI, a startup in which Meta has heavily invested, contributing $14.3 billion to bolster its competitiveness in AI.
The potential move to a closed model symbolizes a significant change in Meta's longstanding commitment to open-source development, a strategy that has garnered support from many in the tech community, including prominent figures such as Yann LeCun, who noted, "the platform that will win will be the open one." That statement leads to interesting discourse about the implications of such a transition amidst an intensifying rivalry with tech giants like Google and OpenAI.
Despite these conversations signalling a shift in direction, the discussions are still in their infancy, and no decisions have been finalized. Any potential change would require ultimate approval from CEO Mark Zuckerberg. This comes amid delays in the release of the Behemoth model, which has seen setbacks due to internal performance issues, prompting a reevaluation of testing protocols within the lab.