Applying Social Network Analysis to the comments sections to understand how online audiences negotiate mediated representations of artificial intelligence
Marie Boran
Modern media consumers no longer passively accept mediated representations of science and technology; internet technologies have given them the ability to reject or negotiate these representations in public and quasi-public digital spaces. Social media platforms and the comments sections of news websites provide the media consumer with an opportunity to publicly react to stories about science, provide feedback to the author, and to engage in discussions with fellow readers.
In this paper, we explore the potential for Social Network Analysis in combination with thematic analysis and discourse analysis to shed light on how participatory audiences make sense of artificial intelligence. There exists some research on media frames and artificial intelligence (AI): a 2018 study from the University of Oxford finds that much coverage in the UK is industry-led while many stories focus on controversial public figures such as Elon Musk, thus framing AI in a particular way for public consumption.
However, there is a dearth of research on public perception and understanding of AI. Our study mines the comments section in order to understand this in more depth. We take into account both the content of individual comments and the comment network below stories on AI in The Guardian online: comments, commenters, their in-degree (replies) and out-degree (comments they reply to).
We compare media frames to the unique frames through which these online audiences are negotiating artificial intelligence (risk; loss/gain; personal) and assess how this is negotiated i.e. the presence or absence of deliberative discourse. Social Network Analysis provides important context to these comments by visualising them as a conversation amongst a commenter community. While multiple studies rate online discourse as low in deliberative quality (and therefore unable to facilitate discussion and dialogue worthy of a virtual public sphere) we take this one step further by using SNA to assess how comments low in deliberative quality are integrated the commenter network overall.
Our findings suggest that particularly negative, uncivil, or inappropriate comments (that are not already removed by the moderator) in relation to AI are largely ignored by the network i.e. these nodes have an in-degree of 0. Meanwhile, comments that have deliberative potential (civility, reciprocity, argumentation, topicality) tend to have a higher in-degree. We also draw upon SNA to explore aspects of these virtual communities including argumentative strategies; individual and group identity; social and political structures; power and prestige relations. By examining the above, we explore the possibility of the comments section as a potential virtual public sphere for discussion of the societal, economic and ethical implications of AI.← Schedule