It is no secret that tech giants like Microsoft, Nvidia and Facebook’s parent company, Meta are in a race to build an AI Supercomputer. Now, a statement put out by Meta suggests that it may be pulling closer to the finish line. In a statement issued, Meta CEO Mark Zuckerberg stated: “Meta has developed what […]
It is no secret that tech giants like Microsoft, Nvidia and Facebook’s parent company, Meta are in a race to build an AI Supercomputer. Now, a statement put out by Meta suggests that it may be pulling closer to the finish line.
In a statement issued, Meta CEO Mark Zuckerberg stated: “Meta has developed what we believe is the world’s fastest AI supercomputer. We’re calling it RSC for AI Research SuperCluster and it’ll be complete later this year.”
It’s not exactly a “supercomputer”
In very simpler terms, Meta’s ‘AI Supercomputer,’ is mostly focused on the application of machine learning. This calls for less amount of accuracy, as compared to the tasks given to a traditional supercomputer. Hence, an AI Supercomputer can carry out more calculations per second, compared to its traditional counterpart.
So, when Meta says that it’s building the “world’s fastest AI supercomputer,” it isn’t necessarily comparing it to the traditional supercomputers scientist or engineers use.
It’ll be Meta’s Big Brother and Gandalf An AI Supercomputer is measured on the performance of graphics-processing chips, which are used for running “deep learning” algorithms. These algorithms help in understanding what’s in an image, analysing a text and translating between languages. Meta’s obvious aim here is to use AI Research SuperCluster (RSC) in a range of systems across its businesses. This includes everything from content moderation and detecting hate speech to augmented reality features on Facebook and Instagram.
In a blog post put out by Kevin Lee and Shubho Sengupta, Meta’s engineers, they said, “RSC will help Meta’s AI researchers build new and better AI models that can learn from trillions of examples; work across hundreds of different languages; seamlessly analyze text, images, and video together; develop new augmented reality tools; and much more.”
So how does this help us?
In the same statement, Meta has made some interesting claims which got our attention. One of the claims says that the RSC can power real-time voice translation to a large group of people, speaking in a different language. Just like the ‘Universal Translator’ in Star Trek.
Additionally, the company has said that it’ll “incorporate real-world examples” from its own system into training its AI. This means Meta will use its customer data to train the artificial intelligence, as compared to the open-source one it’d been using previously.
Would that be a good or a bad thing? We’ll know soon.