Meta Tech

The original blog post would be a better link to post here rather than the article: https://tech.fb.com/artificial-intelligence/2022/07/how-ai-could-help-… Reading through it, I strongly disagree with FB's example for "Better citations in action". I don't see an improvement in the wording and IMO they would be making it worse by switching from an official first party source to a third party one. While their tool might be useful to find semantic (mis)matches, a much more important part of verifying citations is to verify that the source has any business to make claims about the matter in the first place. https://xkcd.com/978/ QUOTE Better citations in action Usually, to develop models like this, the input might be just a sentence or two. We trained our models with complicated statements from Wikipedia, accompanied by full websites that may or may not support the claims. As a result, our models have achieved a leap in performance in terms of detecting the accuracy of citations. For example, our system found a better source for a citation in the Wikipedia article “2017 in Classical Music.” The claim reads: “The Los Angeles Philharmonic announces the appointment of Simon Woods as its next president and chief executive officer, effective 22 January 2018.” The current Wikipedia footnote for this statement links to a press release from the Dallas Symphony Association announcing the appointment of its new president and CEO, also effective…

Closest I am able to think of is recent tool Facebook released for checking citations on Wikipedia: https://tech.fb.com/artificial-intelligence/2022/07/how-ai-could-help-…

I don't know about Peter Jackson specifically but the seperating of different instruments and vocals is a fairly common task, as lots of famous music was recorded on a limited number of tracks, and seperating them out is a part of most modern remixes. I believe there's a couple of open source tools for this, but not sure how they compare with state of the art. edit: article introducing one of them https://tech.fb.com/artificial-intelligence/2020/03/one-track-minds-us…

Reality Labs (Meta) | Technical Program Manager (TPM) | REMOTE or Bay Area | Full-time We are looking for a senior (L5 or L6) TPM to help build next gen tooling, processes, and automation for capacity planning, forecasting, and management. TC: https://www.levels.fyi/comp.html?track=Technical%20Program%20Manager&a… What's in scope: in-house compute (incl. GPU), networking, and storage capacity, public cloud. You can read about our infra here: https://tech.fb.com/engineering/ This is a high impact opportunity. Email resume to malatortsev at fb dot com

Nobody is proposing planting trees as if that makes burning fossil fuels okay. https://tech.fb.com/engineering/2021/04/renewable-energy/ Facebook's renewable energy projects are pretty impressive and if other energy usage estimates in this thread are to be believed they've added more than they consume.

https://tech.fb.com/engineering/2022/03/alaska-terragraph/ supposedly it was developed for remote areas. Hard to tell if there's something special about where it was deployed....maybe the notion of remote has to be still close to wired basestation hubs.

The 70% thing came from a business development official in Loudoun County, a person whose job is to promulgate misleading boosterism. It doesn't make even a sliver of sense to someone who thinks it over for a moment. Cooling is a major data center cost, but it's not 40% any more. All of the major datacenter operators claim a PUE of 1.11 or less. Facebook has been the most open and vocal about how they achieve this, for example at https://tech.fb.com/hyperefficient-data-centers/ (scroll down to "Systems").

You need a swivel chair and a one meter radius of empty space. You need to touch-type, and you can't use a mouse (you point your head instead). You have to learn a system of hand gestures. But if you can adapt to the constraints, a VR workstation allows you to bypass limits like Fitt's Law [1] so you can work "at the speed of thought". Just like vim or emacs, VR is slower and more difficult at first. But in the long run, using it provides an advantage, and so it will be used. Will the future of VR productivity be open-source projects like SimulaVR? [2] Or will it all be part of Facebook's Metaverse? [3] [1] https://en.wikipedia.org/wiki/Fitts%27s_law [2] https://github.com/SimulaVR/Simula [3] https://tech.fb.com/connect-2021-our-vision-for-the-metaverse/

I think this research shows a lot of promise when paired with current work being done at FRL (Reality Labs) with regards to haptics[0][1]. Having overall multisensory integration and proper proprioception will be very important to reaching realistic AR/VR, especially in situations demanding real-time feedback. [0] - https://www.youtube.com/watch?v=v9bbbYcgMqQ [1] - https://tech.fb.com/inside-reality-labs-meet-the-team-thats-bringing-t…

Seems pretty clear companies are talking about "the metaverse" as broad concept. https://tech.fb.com/connect-2021-our-vision-for-the-metaverse/ https://www.epicgames.com/site/en-US/news/announcing-a-1-billion-fundi… https://www.matthewball.vc/about

First hands-on: https://www.google.com/amp/s/www.theverge.com/platform/amp/2021/9/9/22… FB announcement: https://tech.fb.com/ray-ban-and-facebook-introduce-ray-ban-stories-fir… Ray-Ban link: http://www.ray-ban.com/discover-ray-ban-stories

They already are, they bought CTRL-Labs which made a prototype wristband for reading motor neuron activation. They have a demo on YouTube showing it for keyboarding as well as for surprisingly accurate 3D hand pose estimation. CTRL-labs is the one behind https://tech.fb.com/inside-facebook-reality-labs-wrist-based-interacti…

From Facebook Reality Labs: > Today in Nature Communications, Chang and David Moses, a postdoctoral scholar in Chang’s lab at UCSF, published the results of a study demonstrating that brain activity recorded while people speak could be used to almost instantly decode what they were saying into text on a computer screen. While previous decoding work has been done offline, the key contribution in this paper is that the UCSF team was able to decode a small set of full, spoken words and phrases from brain activity in real time — a first in the field of BCI research. The researchers emphasize that their algorithm is so far only capable of recognizing a small set of words and phrases, but ongoing work aims to translate much larger vocabularies with dramatically lower error rate https://tech.fb.com/imagining-a-new-interface-hands-free-communication…

"Forget 3DoF & AR" Sidenote, Quest will be mixing between AR and VR [1]. Maybe it'll take a hardware revision to really bring it to the fore, but the future of AR/VR are quite blurred. [1]: https://tech.fb.com/the-future-of-work-and-the-next-computing-platform…

FB link: https://tech.fb.com/imagining-a-new-interface-hands-free-communication…

Very cool that they are integrating it with HOT tasking manager and making it easy for anyone to use the editor with the ML-generated proposed objects in several countries. I think the ML has been there for a couple years but currently it's not easy to take advantage of it. Hopefully they eventually release the ML pipeline itself as well. Their RapiD editor has some similarities to a research project I was involved with: https://mapster.csail.mit.edu/maid.html (https://www.youtube.com/watch?v=i-6nbuuX6NY vs https://tech.fb.com/wp-content/uploads/2019/07/add_ML_road.gif.-1.gif)