Meta uses 'human level' AI for metaverse

 Meta has just announced Project CAIRaoke, the project to bring AI to "human-level" intelligence with

AIRaoke is basically the same AI tools that Google's DeepMind company is working on. They mainly learn and process situations through machine learning technology.

To enhance its AI capabilities, Meta has invested heavily in machine learning. At the same time, the company started rolling out self-supervised learning features - tools that allow AI to recognize patterns on its own.

For a typical self-learning AI model, including the one Meta is using for content moderation on Facebook, the algorithms work by recognizing labels and tags entered by humans. For example, when Facebook moderators include content labels, the AI ​​will conduct censorship based on those labels.

"So far, Facebook's AI testing for text and video is outperforming previous models," Zuckerberg said. "AI is still in its infancy, but we think it will be an important tool for the metaverse."

Joelle Pineau, who leads Meta's AI research team, says the new system could be used in robotics, including humanoid machines in the future. He also emphasized that the company is testing AI for robots in a real environment to allow them to learn like humans.

"AI will learn faster, deeper, self-transfer knowledge once they are practiced in the real world through steps: trial, error and retry," Pineau said during the event.
Previous Post Next Post


{ads} x