Facebook Monitoring
This article touches on Facebook’s social media content posted and how some of the content is considered to be toxic. These noxious posts revolve from extreme violence, pornography, unpleasant images, and even drug distribution related messages. At first, Facebook hired a third party company which had thousands of employees monitoring every post being published. However, the volume of published posts far exceeds the monitoring rate. An AI was developed in order to help speed the monitoring process. For example if a nude image is posted and is flagged, the post will be taken down within 24 hours. In addition, the third company also removed fake user accounts. 90% of toxic posts were being removed however, the remaining posts that slipped were needed to be checked by actual people. A consequence of this is that employee’s mental health were harmed by viewing hundreds of disgusting posts. In response, the company provided mental health resources to accommodate with heavy trauma.
It is surprising to see how well Facebook is dealing with the overwhelming amount of posts being published and being able to monitor the majority of posts. I believe that this job is necessary for every social platform because if a social platform is unchecked it can be an unsafe environment. I am in full support of the advocacy of greater mental health resources for employees because I personally would go crazy if I had to remove the most toxic messages people come up with in an everyday job.
When I first came across this article, I immediately thought about the networks class I was taking this semester. This article reminded me of the homework problem set 1 which had a prompt asking which node would be considered to be the faker account. Similarly, the facebook monitoring company must bruising from principles from the networks class such as triadic closures. The importance of this concept is probably what they have implied onto the AI that monitors Facebook’s posts. Briefly, one way to check if an account is fake is if that account is messaging the same friend group over time and if they are receiving messages as well. Fake accounts do not have friends that message them. Another concept that is crucial to understand is that friendships evolve over time. This means that a real account will accumulate friends from one friend which would form a triangle. A fake account would not necessarily be forming friends with their friends. For example, a fake account could have a high number of friends but those friends would not be friends with them or the friends would not have a relationship with anyone, meaning that they are only random people.
Source : https://www.nytimes.com/2021/08/31/technology/facebook-accenture-content-moderation.html