Facebook plans to stop drunken pics appearing online


Ok, I’m sure that we’ve all had those nights where were out in the town, having a good night, things get out of hand, then the phone comes out and all these drunken selfies and photos get taken, then the next thing you know, your silly funny selfies have been plastered all over Facebook.

Facebook is now working on developing software that could prevent users from posting unflattering pictures of themselves or others whilst under the influence by asking users if they really want to post the pictures.

By bringing together image recognition and artificial intelligence, the system would be able to analyse photos and distinguish between sober, drunk and potentially embarrassing photos. Facebook is already using image recognition which helps identify faces which allows users to tag other users correctly.

Yann LeCun, Facebook’s chief of artificial intelligence research lab, had spoken to a source from Wired saying, “Imagine that you had an intelligent digital assistant which would mediate your interaction with your friends, and also with content on Facebook”.

The Facebook digital assistant will say and recognize when you are uploading an embarrassing photo of your late night shenanigans. In a virtual way, Yann LeCun explains, that the assistant would ask you: ‘Uh, this is being posted publicly. Are you sure you want your boss and mother to see this?”. In the future, this assistant could also be able to help identify when another user has uploaded a picture of someone else without their permission, he says.


An ‘AI Solution’

The firm is increasingly using the technology in partnership with the artificial intelligence software. The aim for Facebook’s ‘digital assistant’ is that it will analyse every action that a user is making within the social network, beyond the point of photos to videos and interactions, to give them much more control. Such as, notifications are given when other users post unauthorised photos.

“AI is already being used to examine overall Facebook behaviour in order to identify the right content for news feeds” Mr. Lecun told Wired. “You will have a single point of contact to meditate your interaction but also to protect your private information,” LeCun explained. He also added that “you need a machine to really understand people and be able to hold all that data. That is an AI-complete problem”.  AI-complete is the term that is used to describe the most difficult challenges in artificial intelligence research which is equivalent to creating human-level intelligence. This means that it is a long way from becoming reality.

The next stage would to analyse text thats in posted statuses and then automatically suggest relevant hashtags. Mr. LeCun also spoke about a future in where an intelligent digital assistant could ‘mediate your interaction with your friends’. Many big firms have/are experimenting with AI. Google now has its own AI labs and has been investing heavily in AI and robotics firms. LeCun and his team are now looking towards AI systems that can understand Facebook data in much more complex ways.