Mark Zuckerberg announced on Instagram that Meta is working on building “general intelligence” for AI assistants and “open sourcing it responsibly,” and Meta is bringing together its two major research groups to make it happen. Zuckerberg did not specifically mention the phrase “artificial general intelligence,” but it seems like he is steering in that direction. He went on to explain the importance of general intelligence, stating that intelligence has all these different capabilities where you have to be able to reason and have intuition. He suggested that AGI won’t be achieved all at once, but gradually over time. He mentioned that the Meta is amassing a monstrous GPU capacity to support the training and running of AI models, “350,000 Nvidia H100s, or around 600,000 H100 equivalents of compute.”
This announcement from Zuckerberg and other experts promotes a business-as-usual tone about the development of AGI, opposing the world-threatening hype and fear that have surrounded this topic. This calmness and downplay attitude could be good for the development of AI and perhaps some adjustment of expectations is in order as we see that large language models, as interesting as they are, might not be fully ready for widescale reliable use. It seems like Meta is taking steps to responsibly incorporate AI assistants into a wider range of daily activities and to have a plan of action to bring a full AI-integrated metaverse into the public’s daily lives.