Ethical considerations of machine learning based tools in media research: A case study on BBC Media Action’s Klahan-9 project in Cambodia and El Kul in Libya

Abstract

The most frequent applications of artificial intelligence nowadays come, not in the form of sentient robots, but as code and large pools of data. The use of computational methods to analyse large quantities of information is a very helpful tool for a wide range of disciplines, from finance analysis to weather forecasting. When it comes to media research, employing big data analysis can provide great insight into an audience - and thus help create media products targeted at a specific population. Despite its benefits, media researchers must also take a look at what comprises an ethical use of such technologies.This study focuses on analysing the ethical considerations of using big data research, specifically machine learning tools, in media research. The examination of ethics frameworks is applied to two BBC Media Action projects: Klahan-9 in Cambodia and El Kul in Libya. The analysis of both case studies does not only intend to provide a deeper understanding on data ethics applied to media but also aid in the creation of ethics guidelines for media researchers.As the BBC Media Action research team had questions in regards to both the technology and its implications, the study adopted a participatory approach. Workshops were pivotal in providing the researchers with relevant information regarding big data and machine learning while also setting the groundwork for the creation of an ethics guide. These sessions discussed aspects such as informed consent within a big data context, ensuring the privacy and anonymity of the dataset (as well as the participants that comprise the data), the publication of both the results and the data as a means of ensuring transparency and the technical knowledge required on behalf of the research team to successfully carry out all of the objectives.The El Kul and Klahan-9 case studies became an ethics testing ground that generated relevant points for the creation of a new framework for media research. Such products, however, must be understood as an ongoing process that require constant feedback through future research on the subject.