Automated trolling and fake news generation in future social media: Computational and empirical investigations of the threat and its implications

Today’s media is ICT-mediated, and ICT’s role is bound to become even more pervasive in the future. In this project, we investigate mechanisms by which social media discourses may become under attack through artificial intelligence approaches. For example, socially competent conversational agents may participate in discussions disguised as humans, or entire ecosystems of interlinked websites may be generated by automatic means with very little human effort. When fake user profiles and content reach an even higher quality and start to convincingly deceive human users in a large scale, they become a threat on media industry, democracy and ultimately, on societal integrity. The project combines qualitative analysis of a) trolling strategies and b) "fake news" articles with modern machine learning methods for generating naturally sounding synthetic text. The project is conducted at the Department of Computer Science, University of Helsinki.