Meta makes a lot of noise about artificial intelligence
Meta makes a lot of noise about artificial intelligence, Meta/Facebook, the US Internet behemoth, has unleashed a flood of material about its AI efforts, which include translation systems, virtual assistants, and metaverse voice command tools.
Meta/Facebook presented a two-and-a-half-hour presentation called “Inside the Lab: Building for the Metaverse with AI” to show the world that it is making progress on the metaverse, AI, and all the other semi-theoretical stuff it now exclusively talks about.
Meta makes a lot of noise about artificial intelligence and, of course, the metaverse.
In it, CEO Mark Zuckerberg and what looked to be the entire company took turns explaining all of the AI-related projects they’re working on, from metaverse voice commands to diversity and inclusion initiatives.
BuilderBot was hailed as a “new tool for fostering metaverse creativity.” It allows users to summon things and places within virtual areas via voice commands. As a result, you may just say, “Let’s have a table over there,” and it will appear.
Project CAIRaoke, meanwhile, was introduced as a way to improve virtual assistants through the use of ‘conversational AI.’ It’s billed as a ‘end-to-end brain model that can power considerably more intimate and contextual discussions than current systems.’ Project CAIRaoke’s model is apparently being used in another of Meta’s products, Portal, and Meta plans to merge it with augmented and virtual reality devices.Meta makes a lot of noise about artificial intelligence.
“We may envisage that the technologies from Project CAIRaoke will underpin next-generation interaction between people and devices in a few years,” a post about CAIRoke concluded. We expect this sort of communication to become the universal, seamless means for navigation and engagement on devices like VR headsets and AR glasses, similar to how touchscreens supplanted keypads on smartphones. Our present model is a significant step forward, but we still have a long way to go before we can fully realise our vision. Both the progress we’ve accomplished so far and the challenges that lie ahead excite us.”
Other parts of the presentation featured initiatives that appear to be using AI to overcome language barriers in the form of a Universal Speech Translator, which it claims to be the first of its kind.
Then there’s SystemCards, a project aimed at creating tools to better understand how and why AI works. “System Cards can serve as a key step in the process toward helping people comprehend what AI transparency looks like at Meta’s scale,” according to the research. We will continue to identify new pilots to conduct as the industry matures and talks around model documentation and transparency continue, and we will iterate on our approach over time to accommodate product changes, evolving industry standards, and expectations surrounding AI openness.”
TorchRec is a PyTorch domain library for Recommendation Systems that provides ‘common sparsity and parallelism primitives, enabling researchers to develop state-of-the-art customization models and deploy them in production,’ according to the documentation.
Meta also unveiled the Artificial Intelligence Learning Alliance (AILA), a project aimed at “strengthening diversity and increasing fairness in the field of artificial intelligence.” Meta makes a lot of noise about artificial intelligence.
There was a lot more, but the important message is that Meta is pouring a lot of money into what look to be research initiatives in all kinds of AI and metaverse-related fields, and it wants you to know about it. In addition to the presentation, Meta has produced a series of what appear to be essays rather than corporate releases that go into great detail about its goals and progress on each of these projects.
In some ways, this appears to be an unusual approach in and of itself; organisations who legitimately believe they’ve discovered the next great thing could otherwise focus their efforts on keeping it a secret until they’ve created a market-ready product that they can deploy and profit from. Steve Jobs was known for putting on extravagant shows to promote new technology, but the equipment was already loaded into trucks and ready to go when he arrived. He didn’t hire 400 people to talk about every piece of intellectual property they were working on for years before it culminated into something like the iPhone.