Seite - 38 - in Critical Issues in Science, Technology and Society Studies - Conference Proceedings of the 17th STS Conference Graz 2018

STREAM: Digitalisation of Society Trust in algorithms. An empirical study of users’ willingness to change behaviour. CEPERA, Kay Philipp, KONRAD, Julius, WEYER, Johannes Technology Studies Group, TU Dortmund, Germany Introduction As mobile applications have become part of our everyday life, digital society has begun to use algorithmically generated recommendations as a foundation for decision-making. Whether it is choosing the most efficient route on the way to work, taking an umbrella, if the weather-app forecasts rain or adapting a daily workout routine according to a virtual trainer – algorithms influence our actions at an increasing frequency. Our work aims to answer the question, how users’ willingness to change their behaviour due to such recommendations is being constituted. We therefore look at the factors that influence their trust in app providers and privacy. Additionally, we want to identify how trust itself plays a role in this context and in turn influences the willingness to change behaviour. We establish a research model to grasp the matter theoretically. In order to achieve this, we take a big data process model as a starting point and combine the technology acceptance model (TAM) with theories of trust. Subsequently an empirical study is conducted to validate the postulated coherences. The mentioned recommendations are formed by a process of interaction between users and apps, which depends on large amounts of data. This user generated big data is important for the algorithms to compute a “big picture” of the current situation. On this basis, algorithms generate and then submit recommendations to a large number of individual users, who then have the choice to change their behaviour. Such recommendations can mostly be found in apps that use “Behaviour Change Techniques” (BCT) (Conroy et al. 2014). Both Conroy et al. as well as Middelweerd et al. (2014) provide a taxonomy of such techniques that is displayed in Fig. 1. and recurs to the taxonomy established by Abraham and Michie (2008). Although the mentioned papers focus on the health and fitness sector, it becomes clear, that BCT are used throughout many types of apps. This especially applies for the very popular navigation and weather apps. Therefore, we focused our research on these app types as well as on health and fitness apps. 38