Zuckerberg reveals AI projects to power Metaverse

  • Published
Builder BotImage source, Meta
Image caption,
Mark Zuckerberg created a basic virtual world using Builder Bot, commanding the AI to add features such as an island, trees and a beach

Meta founder Mark Zuckerberg has unveiled several ambitious artificial-intelligence projects, describing AI as "the key to unlocking the Metaverse".

In a livestreamed demonstration, he created a basic virtual world - including an island, trees and a beach - using the AI feature Builder Bot.

Mr Zuckerberg also announced a plan to build a universal speech translator.

"The ability to communicate with anyone in any language is a superpower that was dreamt of forever," he said.

Builder Bot was part of Meta's CAIRaoke project to improve AI assistants and allow "AI to see the world from our experience" as people entered virtual reality via headsets or glasses, Mr Zuckerberg said.

Image source, Meta
Image caption,
Mr Zuckerberg promises privacy-preserving and responsible AI

And he promised the AI systems driving Meta's virtual worlds would preserve privacy and be transparent and responsible.

Facebook has been investing in AI for the past 10 years and has one of the world's leading experts, Yann LeCun as its head of AI.

In January it announced that it had built a new AI supercomputer that it aims to be the fastest in the world when completed in mid 2022.

Image source, Meta
Image caption,
Meta's head of AI Yann LeCun is developing systems that rely on self-supervised learning

Facebook changed its corporate name to Meta in October, following revelations from whistleblower Frances Haugen.

At the time, Mr Zuckerberg talked of a wide-ranging Metaverse - accessible via VR headsets, such as the Meta Quest, formerly Oculus -where people could work, play and chat.

And Meta plans to hire 10,000 people in Europe to help build it.

There has been much hype of its potential.

But critics have asked whether big corporations should be allowed to dominate the creation of such worlds - and how safe users would be.

'Unwanted interactions'

One of Facebook's earliest investors, Roger McNamee, told BBC News the company should be prevented from creating a "dystopian" Metaverse, given how its social network had failed to keep user data private or avoid misinformation and hate speech.

Meta chief technology officer Andrew Bosworth admitted the Metaverse would be much harder to moderate than existing digital platforms, especially given its long-term goal of lots of companies interacting in the same space.

But he promised to allow users to control the experience.

And following reports women felt harassed in Meta VR platform Horizon Worlds, the company hastily introduced a feature called Personal Boundary to protect avatars from "unwanted interactions".