Meta is indeed in the AI race, with the regular publication of new generative AI models to its credit, for example. However, the company is now positioning itself among the players diving into artificial general intelligence (AGI).
Last Thursday, Meta CEO Mark Zuckerberg posted a nearly two-minute video on Threads to update the public on the company’s AI efforts, including a look at its long-term goals for building of the AGI.
“Today, I am bringing together Meta’s two AI research efforts [FAIR et GenAI] to support our long-term goals of building general intelligence and making it available and useful to everyone in our daily lives,” Zuckerberg said.
Meta doubles its investment in graphics cards
He also added that the need to build AI of this type has become more important than ever, as this technology will be crucial to powering the next generation of AI services.
This is why Meta is setting up a particularly powerful infrastructure. By the end of the year, the company will have 350,000 Nvidia H100 graphics cards, and a total of 600,000 graphics cards from various manufacturers.
To put that computing power in perspective, Meta and Microsoft were the biggest buyers of Nvidia’s H100 GPUs in 2023, with 150,000 units each, according to analyst firm Omdia. Meta’s new objective, namely 350,000 H100s by the end of the year, therefore represents more than double its investment to date, which is already considerable.
Everything you need to know about AGI
So what is AGI and why is Meta so focused on investing and developing this technology?
Simply put, AGI refers to AI that can perform complex tasks, such as learning, reasoning, and others, as well as humans.
Currently, even the most advanced generative AI models still need human guidance and instructions to complete a task.
AGI helps remove this human instruction, direction, and supervision because the model can operate autonomously.
“To sum it up, we’re just getting started.”
This development is expected to significantly improve how AI can help humans and even what tasks it can perform.
However, AGI’s unprecedented intelligence and ability to mimic human behavior are often why people approach this technology with fear or trepidation.
So far, AGI has remained a long-term goal, but Meta’s recent announcements show that the technology could be developed sooner than expected.
Concluding the video, Mark Zuckerberg also noted that Meta is currently training LLaMa 3, which is the latest version of Meta’s advanced language model, LLaMa 2, and he discussed the creation of other future models.
“The bottom line is, we’re just getting started,” Mr. Zuckerberg concluded.
Source: “ZDNet.com”