This project analyzed the perception of AI in newspaper articles. The research question that is explored is whether AI is perceived differently in different regulatory environments. Due to data availability issues as well as time constraints, the analysis focuses on the newspaper outlet The Guardian in the UK, US and Australia. The data was retrieved from the Guardian API for the years 2000-2022. A very general query was used to filter out 6813 articles broadly concerned with the topic of AI (or topics related to it). The data was analyzed with the use of Natural Language Techniques like Topic Modelling and Sentiment Analysis.
With the limited analysis at hand we can not find a strong difference in the topics that are discussed in the newspaper articles that are concerned with AI. The same finding holds for the sentiments attached to the topics, which are also mainly similar for all three countries. We find a divergence between countries for the topics of "drones" & "facial recognition" whichcome with more negative sentiments in Australia and in the US compared to the UK. We formulate the hypothesis for further research that this might be due to a potential backlash in less regulated environments (where therefore more "scandals" happen), or due to context specific discussions which differ in their details between the countries.
To dive deeper into the project and the details of our findings a presentation is provided: Presentation Slides
Authors: