Meta Shifts Focus to Artificial General Intelligence and Trains Llama 3
Mark Zuckerberg officially makes the pursuit of artificial general intelligence a core mission for Meta as the company trains its advanced Llama 3 model. The company plans to continue its open-source approach for this new technology as long as it remains safe and responsible.
Mark Zuckerberg officially announces that Meta is pursuing artificial general intelligence (AGI) as a primary company mission. This shift away from the Metaverse as the main focus follows years of quiet AGI research within Meta's dedicated departments. By giving this pursuit official status, Zuckerberg signals a major strategic pivot for the social media giant.
As part of this new focus, Meta is actively training Llama 3, the next generation of its large language model series. The upcoming model promises significant improvements in code generation, reasoning, and planning. Rumors indicate that Llama 3 performs at a GPT-4 level, which represents a massive leap forward for the open-source AI community that currently operates mostly at the GPT-3.5 tier.
Meta intends to maintain its partial open-source strategy for these advanced models, provided the release is safe and responsible. Chief AI scientist Yann LeCun strongly supports this open approach, arguing that freely available AI platforms are inevitable and necessary. This stance directly contrasts with competitors like OpenAI, which initially promised open research but now keep their most advanced systems closed.