Facebook: In response to accusations made by former employee Frances Haugen during testimony in the US Congress, Facebook stated that the prevalence of hate speech on the social network has dropped nearly 50% in the past three quarters. The data appear in a statement released this Sunday (17).
In the document signed by Facebook’s vice president of integrity, Guy Rosen, the company defends its measures to combat hate speech, explaining that it is “wrong” to have only the removal of content as a metric. According to the executive, there are other ways to reduce the viewing of such materials on the timeline.
Rosen states that “prevalence is the most important metric to use because it shows how much hate speech is actually seen on Facebook.” For him, technology has had a great impact on reducing the visibility of this type of post, which was 0.05% of the content viewed in the first nine months of 2021.
That’s five views for every 10,000 posts, a drop of approximately 50% compared to the same period last year, according to the company. The social network also said the data is reviewed by a group of independent international experts.
Problems with Facebook Algorithm
Internal documents leaked by The Wall Street Journal yesterday (17) reveal that Artificial Intelligence tools fail to detect and remove harmful content. As per the publication, less than 5% of this material is actually excluded from the platform.
The report also points out that when Facebook’s algorithm is in doubt as to whether certain content violates company policies, it keeps it circulating but reduces its reach. In addition, the social network would not be punishing users who disseminate misinformation.
In this regard, Rosen said that the technology is not “perfect” and is constantly evolving, with the platform teams looking to identify problems and create solutions.