US Internet giant Meta/Facebook has launched a barrage of information on AI projects it is working on, including translation systems, virtual assistants and metaverse voice command tools.
Keen to show the world it is making some progress on driving forward the metaverse, AI, and all the other semi-theoretical things it seems to now exclusively talk about, Meta/Facebook held a two and a half hour presentation called ‘Inside the Lab: Building for the metaverse with AI.’
In it CEO Mark Zuckerberg and what appeared to be the entire rest of the company took turns to explain all the things they are working on around AI, from metaverse voice commands to diversity and inclusion drives.
BuilderBot was held up a as a ‘new tool for fuelling creativity in the metaverse’. It enables voice commands to summon up objects and environments withing virtual spaces. So you can say ‘let’s have a table over there’, and it appears. This video sums it up:
Elsewhere project CAIRaoke was unveiled, which is designed to make virtual assistants better using ‘conversational AI’. It is described as an ‘end-to-end neural model that can power much more personal and contextual conversations than the systems people are familiar with today.’ Apparently the model that resulted from Project CAIRaoke is being used in another of its products, Portal, and Meta aims to integrate it with augmented and virtual reality devices.
A post published in relation to CAIRoke concludes with: “We can imagine that in a few years, the technology from Project CAIRaoke will underlie next-generation interaction between people and devices. On devices like VR headsets and AR glasses, we expect this type of communication to eventually be the ubiquitous, seamless method for navigation and interaction, much as how touchscreens replaced keypads on smartphones. Our current model is an important step forward, but we have more work to do to fully realize this vision. We are excited by both the progress we’ve made so far and the challenges ahead.”
Other parts of the presentation showcased projects which are apparently leveraging AI to tackle language barriers in the form of a Universal Speech Translator, which it claims provides instantaneous speech-to-speech translation across all languages.
Then there’s SystemCards, which is supposed to be about developing tools to better understand how and why AI operates. The literature published in relation to this states: “System Cards can serve as an important step in the journey toward helping people understand what AI transparency looks like at Meta’s scale. As the industry evolves and discussions about model documentation and transparency continue, we will continue to identify other pilots to undertake and iterate on our approach over time, so we can reflect product changes, evolving industry standards, and expectations around AI transparency.”
Elsewhere there is TorchRec, which is apparently a PyTorch domain library for Recommendation Systems, which provides ‘common sparsity and parallelism primitives, enabling researchers to build state-of-the-art personalization models and deploy them in production.’
Meta also announced something called the Artificial Intelligence Learning Alliance (AILA), which is an initiative designed to ‘strengthen diversity and increase equity in the field of artificial intelligence.’
There was much more communicated besides, but the bottom line is Meta is piling a huge amount of resources into what appear to be essentially research projects around all sorts of AI and metaverse type areas, and it wants you to know it. As well as the presentation itself, Meta has now published a series of what look more like essays than corporate announcements which go into exhaustive detail on its ambitions and progress in each of these projects.
This in some ways this feels like an unusual approach in itself – companies that genuinely think they’ve stumbled on the next big thing might otherwise spend some energy on keeping it under wraps until they have completed a market ready product they can deploy and rake in tons of cash with. Steve Jobs was famous for making big elaborate showcases around new tech – but the gear was loaded in trucks and ready to go when he did. He didn’t roll out 400 staff to exhaustively talk about every bit of IP they are working on years before it coalesced into something like the iPhone.
The fact Meta is making such a spectacle of what it’s R&D department is working on, with the detail of the communication reading more like something a university would publish, makes you wonder who it’s trying to convince. It seems desperate to demonstrate that this all this bleeding edge stuff it has bet the farm on – which isn’t really related to the business model that got it to where it is today – is firstly more than just buzzwords and vague ambition, and secondly a good idea in the first place.