Instagram Uses AI To Eradicate Trolls And Offensive Comments

Social media is a cesspool of people who want nothing more than to cause pain and misery to others. While Facebook, Twitter, and Snapchat have done their best to moderate their own platforms, it would seem that Instagram has finally had enough. Bringing out the big guns, the photo-sharing network is using artificial intelligence to eradicate comments left by trolls.

This is not the first time that an online platform tried to use smart computer technology to moderate its comments section. However, trying to create an algorithm that can detect context as opposed to simply banning all naughty words and phrases is a monumental task. As WIRED reports, humans are quite good at telling the difference whereas machines are terrible at it.

By using a new system called DeepText, however, machines can deduce what a series of words mean based on all the other words around it. For example, if the machine detects the word “white” that is followed by the word “flower,” it will most likely consider it harmless. On the other hand, the phrase “white power” has a much higher chance of getting the attention of the machine.

DeepText was actually built by Facebook engineers in-house, which allows the social media platform to scan massive amounts of text. This allows the social network to customize each user’s experience depending on their interests.

This new comments filtering option is being turned on by Instagram as the default setup, The Verge reports. Previous attempts at moderating the comments section required users enabling the filtering option first.

During the launch, English will be the main target of the system, but the social network promised to include other languages in time. Aside from filtering offensive comments, Instagram also revealed its spam filtering system, which has been in place since last year.