Artificial intelligence as a tool in social media content moderation

Abstract

On social media, users engage with each other through consuming and creating user-generated content, the amount of which has increased alongside the growth of the userbase. This presents challenges to social media companies as some uploaded content can be harmful to users and the appeal of the service. Platforms rely on filtering systems and manual labour to remove inadmissible content in the process of moderation. Advancements in the capabilities of artificial intelligence have made it possible to harness the technology for the purposes of moderation. This thesis explores the potential uses of AI in content moderation through features enabled by machine learning. The study was con-ducted as a literature review. It was found that artificial intelligence can detect instances of toxicity and hate speech as well as harmful images and multimedia content through natural language processing and computer vision technologies. The use of AI enables the scaling of moderation systems, the expedient evaluation of content and expanded moderation capabilities in different languages. Limitations in the implementation of automated moderation systems include inaccuracies, lack of contextual awareness, slow pace of adaptation, and concerns surrounding biases, transparency and freedom of expression.