Facebook is in distant progress in artificial intelligence and machine learning and a lot of new ones to build your metaverse version.
At an event for the company, now-known as Meta, CEO Mark Zuckerberg on Wednesday announced Project Caraoke, an effort to bring artificial intelligence to a “human level” of intelligence where it can learn, predict and act alone.
“In the metaverse, we will have to help people navigate between the virtual world & the physical world, and as these virtual worlds will constantly change, AI need to understand the context and learn like humans,” said Zuckerberg.
Facebook is building AI that can learn on its own
The project, mentioned just like the sing-along activity, is sort of precisely like Google’s DeepMind machine-learning subsidiary. To boost its AI capabilities, Facebook is making an investment heavily in machine-learning technology. The company stated it had started to implement self-supervised learning, a tool that permits AI models to understand patterns on its own.
In a typical AI learning model, including those used by Facebook, algorithms learn from tags and labels putted by humans, such as Facebook content moderators. The AI then makes decisions based on these labels. So far, Facebook’s SSL testing outperforms previous AI models for text & video, Zuckerberg said.
“SSL is still in development, but we believe it will be an important tool for the Metaverse,” he added. “Information in the metaverse is too large to be captured by labeled datasets.
For the Facebook metaverse to be what Zuckerberg said it can become, an immersive virtual world filled with useful and fun apps that will be the next version of the mobile internet, it will need technology that’s lesser cumbersome to deploy & and operate. Self-learning artificial intelligence is a step in this direction.
In a separate segment, Joelle Pineau, a managing director for the organization’s AI-research group, stated the organization’s work in robotics, which includes machines ready with human-like skin. Pineau additionally stated the organization was the use of AI to test robots in real-world settings like homes, ostensibly to permit them to learn as human beings do.
“Learning is a lot faster, deeper, and transferable as soon as it is achieved in the real world through trial and error, rinse and repeat,” Pineau said.
Project Caraoke may improve Facebook’s language capabilities
Caraoke is even a part of the company’s plans to expand its language capabilities, which Zuckerberg described as “a fundamental problem” for the metaverse.
Two projects were announced at the event directed at helping Facebook improve its language capabilities: No Language Left Behind, an artificial intelligence-based translation system for the internet that is expected to learn all languages without using English as a translation base, and then Universal Speech Translator, a tool to translate information and conversations into any language in an augmented reality or virtual reality experience. None have been implemented.
All of the executives speaking at the event acknowledged that the level this AI technology needs to reach for Facebook’s Metaverse ambitions to come true, be it in the virtual or physical world, is at least a few years away.
Meta plans on hiring more AI researchers to build the metaverse
At the end of the event, Zuckerberg and other executives openly asked people to apply to work for Meta in AI and other departments working in the metaverse. The company has lately struggled to retain and recruit talent due to its many issues with malicious content, political misinformation and other social ills.
But even though Meta opens up jobs for AI scientists, self-learning algorithms still have a long way to go before they become useful to company’s business.
As Yoshua Bengio, an AI expert put it succinctly at the conference: “We are still a long way from AI at the human level.”