Instagram will use machine learning to demote 'inappropriate' posts

In the wake of the criticism Facebook has faced in its handling of hate speech, the social media giant is now tweaking features on Instagram to control inappropriate or abusive content.

The Facebook-owned social networking platform is reworking its algorithms to limit the exposure of posts it considers inappropriate. Instagram stated, “We have begun reducing the spread of posts that are inappropriate but do not go against Instagram’s Community Guidelines.”

While Instagram hasn’t provided a proper definition of a post it deems inappropriate, TechCrunch claims that posts with violent, graphic/shocking, sexually suggestive, misinformation and spam content could fall under this category. This could affect sexually suggestive content as well, which don’t actually depict nudity but imply. Content that is considered in bad taste, lewd, violent or hurtful could also see a sharp decline in views.

Instagram content moderators are already undergoing training to begin flagging borderline content with AI as the driving factor behind determining which posts should be recommended.  Instagram’s Product Lead for Discovery Will Ruben said: “We’ve started to use machine learning to determine if the actual media posted is eligible to be recommended to our community.”

Instagram blocking posts it deems inappropriate may have serious repercussions on independent content creators as these posts will no longer appear to the broader community in Explore or hashtag pages. The news has already been met with mixed reactions with several content creators concerned their viewership will start to diminish, inevitably limiting their ability to earn revenue through monetised posts.

At the current moment, the content demotions won’t be extended to Instagram’s feed and Stories, but this could change in the future. There are also questions on the exact classification of how content will be deemed inappropriate. Unlike Facebook, Instagram to a vast extent has managed to keep the lid on inappropriate content, but AI algorithms could end up demoting too much content rather than rooting out the explicitly harmful posts.