Current XAI research shows that explanations of AI need to be tailored to the individual explainee. We investigate whether XAI explanations can be successfully adapted to humans, when based on an appropriate static partner model representing relevant features. More specifically, we analyze the effects of static socio-demographic and individual factors on the advice-taking of different explanation strategies in a human-agent-interaction scenario. Results showed significant effects of the participant’s risk value, the mathematical self-assessment and the distance and direction of the advice to the first selection on advice-taking. Leveraging these results for an adaptation scheme, we train a classifier to predict a suitable explanation strategy based on all static features. We compare this classifier to results from a classifier working on dynamic features. Our results show that dynamic factors are as important as the static ones. Using static and dynamic factors increased the classifier’s accuracy, but the model showed overfitting and no generalization. Dividing the dataset by nationality yielded better generalization performance, indicating that nationality has an effect on predicting advice-taking. In addition, we propose an adapted measurement for advice-taking that considers the adaption beyond the given advice in advice-taking.

This study investigates assistant nurses’ views on and needs for orientation to care robot use in three European countries. The use of care robots is gradually being incorporated into welfare services. Orientation to care robot use (in short, introduction to the use of the care robot technology) has thus become a key issue for care services. A survey was sent to assistant nurses in Finland, Germany, and Sweden, to which 302 participants responded (Finland n = 117; Germany n = 73; Sweden n = 112). Only 11.3% of assistant nurses had experience of giving orientation to care robot use to older adults or colleagues, but over 50% were willing to do so. Those with experience of using care robots should take part in orientation. Orientation to care robot use should be seen as part of care management and an issue that may affect the whole organisation. Management should, firstly, allow assistant nurses to get to know care robots by offering information, and secondly, consider with the assistant nurses the ways care robots can change their work and the implications of this change. Emphasising the social factors and practical orientation to care robot use extends the previous theories and perspectives of technology acceptance, adoption and diffusion.

The rapid demographic shift toward a greater percentage of the elderly population increases the need for welfare services. Welfare technology and especially care robots can be regarded as an important measure to counteract such demographic challenges. However, when implementing new technologies, structured information is of immense importance to develop societal trust. Frequently, research addresses trust solely at the level of the end-user. However, trust at the level of opinion leaders and political decision-makers is also relevant as they are catalysts for trust. This study aims to detect the perceived trust level of users from the viewpoint of opinion leaders (politicians, insurance organizations, and media) in the Swedish, Finnish, and German society. Furthermore, this study uses qualitative expert interviews and identifies four trust categories: trust in the health care system, trust in regulations, trust in technology, and interpersonal trust. The findings stress that targeting only the end-users is not sufficient for developing technology trust in society.

We analyze how traffic congestion affects the usage of cruise control in real-traffic situations. Usually, a negative relationship is assumed, but empirical evidence in naturalistic settings is rare. We make use of a large sample of truck drivers ($N$= 562) in Germany. We take advantage of the volatile traffic density imposed by the first implementation of COVID-19-related measures in Germany and analyze truck drivers resulting application of cruise control using random effects models. We match official traffic density information and the share of cruise control usage as recorded by the telematics information of each truck per day. We find that on average traffic density does have the expected effect: more traffic leads to a lower usage of cruise control. However, we find great heterogeneity among drivers and a strong tendency to react nonlinear to changes in traffic density. Additionally, comparing traffic types, heavy-goods vehicle traffic density has a stronger impact on drivers' usage of cruise control than private traffic density. Again, not all drivers react similarly: we investigate eight different reaction patterns to varying types of congestion. Our results show that even in a favorable environment such as low traffic congestion, the use of cruise control cannot be taken for granted: about 25%–34% of the drivers (depending on traffic type) do not respond to changes in traffic density at all. Our results may help to inform models of traffic flow and traffic automation and may serve as a premise for more realistic assumptions about human behavior in traffic.