Facebook is already one of the most powerful and popular web and app platforms, and it has been for more than a decade. The service, whose beginnings we’re all at least somewhat familiar with at this point, has consistently dominated the world of social media, and based on its constant updates, additions, and acquisitions, it’s clear that it aspires to be more than that. The Facebook of 2017 is more than a space to share inane, personal details about the happenings of one’s day, as was the case 10 years ago. Now, users can share events, buy a car, send money, reserve movie tickets, ask for recommendations, and even broadcast live to billions of people around the world.
Yet, Facebook still isn’t done. In a recent blog the company revealed its latest endeavor includes a keener focus on artificial intelligence (AI) and machine learning. This isn’t Facebook’s first time using AI. The company uses bots for their robust, separate messaging system, and FBLearner which is responsible for ranking and positioning content in your newsfeed, is an example of machine learning. Similarly, Facebook’s latest platform, Lumos, will be a powerful tool for images.
Lumos is different than Facebook’s current image capabilities, which does a pretty good job at facial recognition, allowing you to tag friends pretty easily upon uploading. The goal for the platform is to scan images and videos and analyze the content within them, with the capability of describing the media “like you would to a friend.”
Why Is This Important?
With better alt text capabilities, Facebook can position its service so that users can use text to search for photos and be presented, quite accurately, with a plethora of resources. Additionally, Facebook will be able to recommend visual content to you based on the subject matter within them, not necessarily because they’re popular within your network. It makes Facebook more helpful, in a way.
Another component for which Facebook could and plans to use Lumos is for easily identifying objectionable content. A site that features billions of users is bound to have more than its fair share of media that would be considered offensive. Facebook’s current process of responding to flagged content is not lacking, but automated software could do well to improve that process, or at least make it more efficient.
For more details about Lumos and a few early examples of the platform in action, check out ZDNet’s recent blog on the subject.
Follow @YaserKhalifa12 on Twitter for updates related to tech, machine learning and evolutionary algorithms.