Meta is coming up with their first conference detailing their progress of the metaverse after the major rebranding from Facebook. The virtual event ‘Inside the lab: Building for the metaverse with AI’ will be held on February 23 at 10:30 PM (IST) and have speakers across the AI board. The opening and closing notes will be from the CEO Mark Zuckerberg, where he is expected to speak about the company’s current breakthrough research and future big bets.
Other than Mark’s session, an important session to not miss is expected to be the ‘Toward Self-Learning Vision Systems’ by Piotr Dollar. There is also expected a detailed view on the privacy initiatives by Jacqueline Pan and Stevie Bergman in the session ‘Building Responsible AI Meta.’
Sign up for your weekly dose of what's up in emerging technology.
Other than the company updates, the community is looking ahead to the discussion between Yann LeCun, Lex Fridman, and Yoshua Bengio on AI research and the path to human-level intelligence. They will discuss this future, available research, risks, and key areas of development that will influence our understanding and application of autonomous AI systems. Developers are also looking ahead to Irina Kofman’s session on Meta AI’s approach to open science, industry collaboration, and investments in building and developing a more diverse AI workforce.
All the speakers, including Zuckerberg, are expected to announce developments and future plans. Some of these that we might think will happen are:
Piotr Dollar’s session revolves around the recent breakthroughs in self-supervised learning and how they’re changing the way Meta developers train their computer vision models. It was last month when Meta announced the ‘data2vec,’ a high-performance self-supervised algorithm that works for multiple modalities. In March last year, the company first announced their work on self-supervised learning and called it ‘The dark matter of intelligence.’ The company is highly likely pursuing self-supervised learning quite closely, and an announcement on innovation in this domain is expected. The event might include details of more developments and even increased use cases.
Meta wants to be known as ‘the metaverse’ company, and it has left no stone unturned in terms of marketing it. In terms of research too, Joelle Pineau and Jérôme Pesenti will discuss the immersive future and the advancements in technology. From allowing more avatars to introducing Horizon worlds, 2022 has been the year where Meta is opening up their immersive experience to the world. The event is expected to have announcements of Horizon World’s going beyond the US and Canada. The team is also expected to promote the fact that the Oculus Quest is now ‘Meta Quest’ and might introduce a new device or even new features. Early this year, Meta also introduced AI Research SuperCluster (RSC) – the world’s fastest AI supercomputer, where they claimed that the work with RSC will build technologies for the next computing platform, which is the metaverse. The event may have announcements on RSC’s applications and products.
The USP of Meta’s algorithm of self-supervised learning is that it works for multiple modalities. Meta has been pushing personal assistants, and Meta might announce a personal assistant like an Alexa or a Siri. Meta is conducting research in this direction as it can be guessed from Alborz Geramifard’s session on personalized assistants and the session by Angela Fan on Speech and Translation. In fact, her session details state that they ‘will pull the curtain back on an ambitious body of work that aims to provide people with the opportunity to access anything on the internet in their native language and speak to anyone, regardless of their language preferences.’ This month Meta also added thousands of expert human speech transcriptions to their open-source Casual Conversations data set.
Building responsible AI has been an important initiative among AI companies globally. Also, as Facebook has always been targeted for its loose privacy policies, it is likely that there will be sessions detailing their concerns. Jacqueline Pan and Stevie Bergman’s session revolves around the company’s commitment to building AI responsibly and the cross-disciplinary team supporting these efforts. At the end of 2021, Meta has introduced the AI technology, Few-Shot Learner (FSL), that can adapt and take action on new or evolving types of harmful content in weeks. Early this year, too Meta announced Privacy Center, where users can learn more about their approach to privacy across our apps and technologies. Privacy in the metaverse is also a major concern. Currently, Meta has only announced a ‘Personal Boundary’ that prevents anyone from invading the avatar’s personal space. This event is also expected to detail more of the company’s further initiatives.
This year Meta AI collaborated with Carnegie Mellon University to find efficient and scalable ways to store and use renewable energy. There could be some more announcements on the company going green and planning better ways to manage energy as metaverse is expected to require bigger and stronger data centres. The ideas of virtual lands are coming true with the development of the metaverse. As Meta wants to be the torchbearer of this future technology, it is essential to attend this session and see the way how this tech giant is betting on this one technology.