Democracy Gone Astray

Democracy, being a human construct, needs to be thought of as directionality rather than an object. As such, to understand it requires not so much a description of existing structures and/or other related phenomena but a declaration of intentionality.
This blog aims at creating labeled lists of published infringements of such intentionality, of points in time where democracy strays from its intended directionality. In addition to outright infringements, this blog also collects important contemporary information and/or discussions that impact our socio-political landscape.

All the posts here were published in the electronic media – main-stream as well as fringe, and maintain links to the original texts.

[NOTE: Due to changes I haven't caught on time in the blogging software, all of the 'Original Article' links were nullified between September 11, 2012 and December 11, 2012. My apologies.]

Saturday, February 01, 2014

Facebook's Plan For Artificial Intelligence: Transcribe Your Calls, Decipher Your Photos

On Facebook's earnings call Wednesday afternoon, Mark Zuckerberg offered a peek at the social network's long-term plans for artificial intelligence. And just as we explained in November, Facebook hopes AI will help it more thoroughly understand the meaning of everything you share, from gauging your mood by the words in your status update, to picking out a Coke can in your photos.

Facebook has been working to expand its artificial intelligence research lab, and last month appointed a renowned researcher with expertise in deep learning to oversee it. Deep learning is a sub-field within AI that focuses on training computers to make sense of the many messy, undefined and irregular types of data we humans generate, such as when we speak, write, photograph or film. (Teaching a computer to recognize a cat, for example, turns out to be an extremely difficult problem.)

So what would Facebook do with deep learning capabilities? Get to know you much, much better by more effectively analyzing every item you share, according to Zuckerberg's clues.

"The goal really is just to try to understand how everything on Facebook is connected by understanding what the posts that people write mean and the content that's in the photos and videos that people are sharing," Zuckerberg explained to analysts and investors on the call. "The real value will be if we can understand the meaning of all the content that people are sharing, we can provide much more relevant experiences in everything we do."

Let's try to unpack some of those generalities. In a sense, you can imagine a deep learning-enhanced Facebook as a jealous ex who stalks and overanalyzes your every online move. Instead of merely knowing you'd shared a photo, Facebook might be able to figure out that the snapshot showed a beach, along with a picture of your ex-boyfriend and that you two were smiling. When, a few days later, you post a status update, Facebook could perhaps analyze your phrasing to guess that you're lonely and depressed. And before long, you might be seeing ads for dating sites, antidepressants and funny films.

Zuckerberg also mentioned that AI could be used to transcribe the voice clips people share in Messenger, so people could receive them more easily.

He acknowledged that these are "pretty big tasks in AI" that Facebook's teams are working on, and noted it could be years before they're fully formed. Within the next three years, Facebook will be focused on "building new experiences for sharing," Zuckerberg said. Five years from now, he predicted, we should see Facebook's AI work reshaping our experience.

Original Article
Source: huffingtonpost.com/
Author:  Bianca Bosker

No comments:

Post a Comment