Forget the metaverse (or almost): Meta wants to realize theartificial general intelligence (AGI). The announcement comes directly from the CEO of Meta, Mark Zuckerbergthrough an Instagram Reel in which he explains that the company wants to combine its AI efforts for an open source project.
Meta focuses on general artificial intelligence, according to Zuckerberg
Zuckerberg revealed that the company is working on AGI, i.e.human-like artificial intelligence. To pursue this goal, it has combined two of Meta’s leading AI research teams – FAIR and GenAI. And the goal is to make the project open source.
In the video, Zuckerberg explains that the goal is to create general intelligence, then make it accessible in a responsible way and spread it widely. The CEO further explained that Meta is currently training its next-generation model called Llama 3.
Additionally, Meta is building a huge computing infrastructure to support the future roadmap. It has in fact purchased computational capacity equivalent to 350,000 H100s (Nvidia’s very powerful GPU, excellent for artificial intelligence) by the end of the year. It then reaches almost the computational capacity equivalent to 600,000 H100s.
Zuckerberg then explained that the company is already putting AI into several devices, such as smart glasses Ray Ban Meta. But between the AI that answers users’ questions and the one that thinks like a person, promised by Meta, there is a gulf.
Meta’s chief scientist, Yann LeCun, last December expressed doubts about the arrival of the artificial superintelligence. LeCun had explained that in the next few years we will at most be able to reach the intelligence of a dog or a cat, not that of a human. So perhaps when Zuckerberg talks about “long-term vision,” he means the next few decades.
Stay updated by following us on Google News!
Follow!