Automated media: key challenges and concepts for reception studies

Jacques, Jerry [UCL] Descampe, Antonin [UCL] Claes, Arnaud [UCL] Wiard, Victor [USL-B]

The availability of large amounts and various types of data combined with recent advances in machine learning are increasingly used as a basis for automated decision-making to address complex problems in numerous contexts. In the media field, we observe that these technologies are able to (partially) automate each step of the information production workflow: collection and curation (e.g., story discovery, datamining, social media monitoring), production (e.g., content generation, video editing, data summarization), or distribution (e.g. conversational bots, personalised recommendation). These new techniques can be qualified as "automated media" as they have the particularity to organise and represent reality on the basis of rules established by their creators, but without their direct intervention or supervision. This major evolution triggers the development of new practices for both media workers and audiences. In this context, the objective of this ongoing research is to question the growing automation of our contemporary media environment by developing a theoretical framework for analysing how automation is perceived by users and how it affects their uses of media. This framework aims to better understand how the design given to these automated systems influence (1) users’ perceptions of their potentialities and (2) users' capacities to adopt a critical stance on automated media. To address these issues, we reviewed the scientific literature (e.g. Diakopoulos, 2019; Gehl & Bakardjieva, 2017) in order to identify the different approaches existing on the topic of automated media. The result is the identification of two key concepts relevant for approaching the complexity of automated media and to study how users interact with them : incarnation and virtuality. We propose to define incarnation as the concretisation of a technical system into a defined set of representations and affordances designed to influence users’ mental models by leveraging characteristics usually attributed to humans (e.g., appearance, behaviour, use of speech, reasoning capabilities). Depending on the complexity of incarnation, users may be more or less inclined to speculate on the capabilities of the system and to interrogate the intent of their designers. The incarnation of automated media is also deeply related to their level of virtuality (Lévy, 1998), their capacity to offer a wide range of interaction opportunities that may mimic the complexity of human interactions. This theoretical framework may encourage the discussion within the scientific community on the impact of automated media. It also lays the ground for future empirical studies on their reception by audiences.