In an exclusive interview, President Trump went on-record to warn news and reporting site Reuters of the “potential dangers” he believes are associated with the banning of Facebook and Twitter accounts.
Also according to the article, Trump had previously used his Twitter profile, which has amassed over 52 million followers and has become an integral part of his campaign, to launch a series of tweets accusing unnamed companies of silencing a number of right-wing / Conservative voices, “some good and some bad”. Although Trump did not name the companies responsible or provide evidence to his claims, he did suggest that the bias being displayed here was an act of discrimination against conservative platforms and individuals.
“I won’t mention names but when they take certain people off of Twitter or Facebook and they’re making that decision, that is really a dangerous thing because that could be you tomorrow,” read one Tweet.
And while the companies were never named in the tweets themselves, they did closely follow the action by Apple, Youtube and Spotify to ban Conservative News Platform Infowars from their sites, so it’s not too difficult to make an educated guess on who the president is referring to.
If one was to ask the companies, however, these acts would not be examples of discrimination but acts of the companies behaving in obedience with their own policies. Facebook, for example, claimed the ban was due to Infowars’ violation of its rules on hate speech and bullying, and claimed that one of the statements made by the platform actively called for violence against another group.
Twitter and Facebook have declined to make an official statement on Trump’s tweets and accusations. Google and Apple did not respond to the request for a comment.
And although Trump did not feel the need to supply any of his own evidence in his Tweets, there is plenty of evidence elsewhere to support his claims. For example, some former Twitter employees admitted that Twitter uses a “shadow banning” algorithm to remove users from its conversation without telling them they have been removed. In addition, one of the employees admitted that the algorithm bans accounts based on word patterns, and that the targeting is set up so most of the accounts affected by these bans will be republican accounts.
A lot has been happening concerning the topic of Infowars and Alex Jones recently, and the topic has been obscuring many news stations over the past few weeks. Infowars and many of the statements made by the platform were aggressive and offensive to many, so many of its tweets and posts may have, in different ways, violated the anti-bullying and anti-hate speech policies designated by many social media platforms. Which begs the question – Why hadn’t this platform been banned a long time ago?
Using an AI based on word recognition has undoubtedly been tested in the past, and most likely would end up banning a large number of accounts that posted the comments ironically or as jokes. In addition, giving an AI that much “policing power” on a social media site would leave users in constant fear of being banned, which would diminish the platform’s viability for real and open discussion.
So the solution appears to be to use humans to read through comments and pick out obvious “hate speech”. But this has the opposite problem: Humans would accomplish this task at a much slower rate than an AI, leading to many examples of “hate speech” and / or “bullying” to slip through the cracks.
And although Facebook and Google have not revealed what exact methods they use for banning / suspending accounts, it might be some combination of the two. Perhaps, for examples of word strings that reflect obvious hate speech, such as “I hate f*gs”, or other such language, the account is instantly banned, but for other examples, the AI tags them as “questionable” and puts them in a list for human employees to assess in more detail later.
And here’s where the central problem being experienced by many begins to rear its ugly head: Humans aren’t absolute. What one views as hate speech may be seen as perfectly acceptable by another. This causes a level of leeway given to certain accounts, and not to others, that makes the “rules” of such social media accounts seem unfair or seem to reflect a personal bias.
This is why one can point at a post that says a discriminatory left-leaning phrase that was let through the system, then to a post of a discriminatory right-leaning phrase that resulted in an account ban.
And although this bias may not be intentional, if most social media employees side with a certain political demographic, they may be more inclined to mark opposing political views as “discriminatory” due to subconscious reactions.
But a true solution to this issue is not clear. A way for people to police themselves would only worsen the issue, as the dominant political side would quickly overtake the less-represented side. In addition, a poorly-designed AI would result in social media sites that feel more like an overpopulated prison than an open discussion table.
One thing’s for certain, however: If news stories continue to focus on these censorships and biases in social media accounts, platforms will need to start finding better solutions to these problems or they could face serious economic issues.