Voice assistants are becoming ever more pervasive. When it comes down to it, typing, especially on smaller, less wieldy mobile screens, isn’t exactly the easiest way to input commands or communicate with others. It requires a lot of concentration and more than a little dexterity. Recognizing human speech correctly is one of the greater accomplishments of the modern digital era — parsing and understanding human meaning from that speech remains the elusive golden egg. Get that right, and your voice assistant changes U/I and U/X quite literally forever. It would transform consumer use cases over night. And while Siri, Google Assistant, Alexa and Cortana all are impressive platforms, one of the leaders in artificial intelligence may be joining the fray in a major way.
According to Business Insider’s Alex Heath, Facebook may have tipped its hand a bit, unintentionally:
An unreleased voice search feature was recently uncovered in code from a past version of Facebook’s mobile app by developer Blake Tsuzaki, who shared his findings with Business Insider.
The code, which was uncovered by Tsuzaki and first surfaced by The Next Web’s Matt Navarra on Twitter last week, shows what is evidently an unreleased voice assistant capable of making suggestions and finding basic information like the score of a sports game.
The voice assistant was first spotted in a September version of Facebook’s mobile app, Tsuzaki told BI. “It’s non-functional right now (listens, and then errors out), but I’m finding a way to get it to do a bit more,” he said via email.
Why would Facebook make such an investment of time and resources to something they’re already well behind in?
For one, Facebook is one of the most interacted with websites/platforms/digital destinations in the world. And, more and more people have come to access it via mobile, favoring that over any other screen type. As such, modernizing the interface to allow users to interact with the app more naturally, fluidly and efficiently makes perfect sense.
High grade voice assistants also require cutting-edge artificial intelligence to parse human speech, make sense of it, and return a sensical, relevant, useful answer. You can’t just program the software to recognize specific speech strings and then write coded instructions for precisely how to respond to that one, minute use case — you’d be coding forever. There’s no possible way you could get to every iteration of question or expected answer one could hope into the software’s framework. No, instead, it’s far more useful and successful to provide neural networks with reams of human-to-human data for the computers to crunch, identify trends and make sense of. That way, the machine can learn from real human conversations and get to a place where its answers mimic the human answers the deep learning engine has grown to emulate.
Another reason Facebook would be interested in moving consumers to a voice interface? Because it already has all your data, and human data is precisely what’s needed to power A.I.-backed voice assistants. Facebook is at the perfect nexus between platform and service to capitalize on a shift to voice interfacing. The company has unending beau-coups of data that we have provided to it, free of charge I might add, for years. It’s in the ideal position to put that data to work by training neural networks on precisely that data. In a lot of ways, it might be the most ready to take a voice assistant to next-level functionality given Facebook’s investments in artificial intelligence as well as its preexisting data trove:
Artificial intelligence has become a vital part of scaling Facebook. It’s already being used to recognize the faces of your friends in photographs, and curate your newsfeed. DeepText, an engine for reading text that was unveiled last week, can understand “with near-human accuracy” the content in thousands of posts per second, in more than 20 different languages. Soon, the text will be translated into a dozen different languages, automatically. Facebook is working on recognizing your voice and identifying people inside of videos so that you can fast forward to the moment when your friend walks into view. — Fast Company
It’s no surprise that Facebook would want to extend that A.I. reach into both greater data collection possibilities as well as real-world use cases for the tech it’s building. The company is maintaining that a voice assistant is not on the docket any time soon, but the Business Article author seems skeptical:
The company signaled earlier this year that it wasn’t working on a voice assistant when Messenger chief David Marcus told Variety, “We are not working on that actively right now.”
But people familiar with the matter recently told BI that Facebook is, in fact, working on a voice assistant within Building 8, its mysterious consumer hardware division that was just recently formed. Facebook’s assistant is intended to power the company’s forthcoming video chat device for the home codenamed Aloha, the people said.
Aloha is scheduled to make its official debut in May 2018 and compete with the $229 Amazon Echo Show, which was released in June.
Facebook undoubtedly has big plans for storming into the 2020’s at the head of an artificial intelligence armada; a voice assistant may just be the first mate it needs to succeed in doing so.