Meta is home to one of the world’s preeminent artificial intelligence (AI) research labs. It is now treading a very difficult path in the burgeoning world of AI. The company has rapidly established itself as an AI development powerhouse. It is failing to successfully bring those innovations to market and is lagging behind organizations like OpenAI, Anthropic, Google DeepMind, and xAI. This troubling dynamic has us concerned for the longterm direction of Meta’s commitment to openness in AI.
Mark Zuckerberg, CEO of Meta, has been very vocal in promoting the Llama family of models. He points to their open-source nature as their main competitive edge. The Llama Extended family is now open to researchers and developers alike! It creates an ecosystem that encourages cooperation, collaboration and ultimately innovation across the much larger AI community. Zuckerberg’s comments in favor of a more cautious approach are troubling.
In fact, just a few weeks ago in highly publicized meetings, Zuckerberg took OpenAI to task for going more closed after its Microsoft partnership. One area of progress, he said, was Meta’s commitment to transparency. This position is a world apart from recent actions taken and demonstrates a clear commitment to ensure advanced AI research continues in the open.
To power all of this, Meta is doubling down on its ambitious goals by pouring billions into constructing new, energy-hungry data centers. These facilities are the backbone of creating cutting edge AI technologies, such as artificial general intelligence or AGI, otherwise known as “superintelligence.” My company is poised to meet the daunting financial challenges of these complex improvements. This underscores their deep dedication to pushing the frontier in AI, sustained over many years.
Zuckerberg’s recent remarks have introduced an element of uncertainty regarding Meta’s future openness. He has been a vocal proponent for releasing the best in class open-source models. At the same time, he acknowledged that not all progress would be public. His statement is a welcome indication that he has a nuanced understanding of the responsibilities that come with the release of advanced AI technologies.
“We’re obviously very pro open source, but I haven’t committed to releasing every single thing that we do. I’m basically very inclined to think that open sourcing is going to be good for the community and also good for us because we’ll benefit from the innovations. If at some point, however, there’s some qualitative change in what the thing is capable of, and we feel like it’s not responsible to open source it, then we won’t. It’s all very difficult to predict.” – Mark Zuckerberg
At Meta, we’re big believers in the power of collaboration and innovation through open-source projects. It puts a high premium on being careful with technologies that are either dangerous or transformative (and potentially both). How well the company can continue to balance these conflicting concerns will likely determine its standing and sway within the AI community going forward.
A spokesperson for Meta affirmed the company’s ongoing commitment to openness, stating, “We plan to continue releasing leading open source models.” However, with Zuckerberg’s recent comments indicating a shift in mindset, stakeholders within the tech industry are left contemplating what this might mean for the future of Meta’s contributions to AI research.