Today, many families are finding themselves with the option to have their very own assistant at home in the form of a smart speaker. To meet the growing scientific need to better understand how families with young children use these devices at home, we showcase ongoing work that proposes a revised conceptualization of families' smart speaker use based on three important dimensions: how often the smart speaker is used (i.e., frequency), for what it is used (i.e., purpose), and by whom it is used (i.e., form) – so by the parent, the child, or by both together. For the assessment of families' smart speaker use, we combine self-reports of 400 Dutch parents with data donations of their smart speaker log-histories. With this, we not only test for the feasibility of this novel approach under the framework of the GDPR in the EU, but to also draw methodological conclusions about the alignment between self-reported and observed measures of smart speaker use. In addition, we offer an updated typology of different use-patterns across families that can further aid the study of predictors, conditions, and consequences of smart speaker use. Learning from first pilot data and the challenges we have faced, we offer preliminary conclusions and provide suggestions for future work.

Conversational recommender agents are artificially intelligent recommender systems that provide users with individually-tailored recommendations by targeting individual needs and communicating in a flowing dialogue. These are widely available online, communicating with users while demonstrating human-like (anthropomorphic) social cues. Nevertheless, little is known about the effect of their anthropomorphic cues on users’ resistance to the system and recommendations. Accordingly, this study examined the extent to which conversational recommender agents’ anthropomorphic cues and the type of recommendations provided (user-initiated and system-initiated) influenced users’ perceptions of control, trustworthiness, and the risk of using the platform. The study assessed how these perceptions, in turn, influence users’ adherence to the recommendations. An online experiment was conducted among users with conversational recommender agents and web recommender platforms that provided user-initiated or system-initiated restaurant recommendations. The results entail that user-initiated recommendations, compared to system-initiated, are less likely to affect users’ resistance to the system and are more likely to affect their adherence to the recommendations provided. Furthermore, the study’s findings suggest that these effects are amplified for conversational recommender agents, demonstrating anthropomorphic cues, in contrast to traditional systems as web recommender platforms.

Conversational recommender systems provide users with individually tailored recommendations in a flowing dialogue. These require users to disclose information proactively or reactively for receiving personalized recommendations, which can trigger users' resistance to the platform and to the recommendations. Accordingly, this study examined the extent to which user-initiated and system-initiated recommendations provided by a conversational recommender system influenced users' perceptions of it. The results of an online experiment entail that when recommendations are system-initiated, as compared to user-initiated, users perceive to be in less control and perceive the system as riskier. Furthermore, the results stress that systems that provide user-initiated or system-initiated recommendations do not differ in users' perceptions of anthropomorphism.

The ongoing substitution of human decision makers by automated decision-making (ADM) systems in a whole range of areas raises the question of whether and, if so, under which conditions ADM is acceptable and fair. So far, this debate has been primarily led by academics, civil society, technology developers and members of the expert groups tasked to develop ethical guidelines for ADM. Ultimately, however, ADM affects citizens, who will live with, act upon and ultimately have to accept the authority of ADM systems. The paper aims to contribute to this larger debate by providing deeper insights into the question of whether, and if so, why and under which conditions, citizens are inclined to accept ADM as fair. The results of a survey (N = 958) with a representative sample of the Dutch adult population, show that most respondents assume that AI-driven ADM systems are fairer than human decision-makers. A more nuanced view emerges from an analysis of the responses, with emotions, expectations about AI being data- and calculation-driven, as well as the role of the programmer – among other dimensions – being cited as reasons for (un)fairness by AI or humans. Individual characteristics such as age and education level influenced not only perceptions about AI fairness, but also the reasons provided for such perceptions. The paper concludes with a normative assessment of the findings and suggestions for the future debate and research.

The increasing volume of images published digitally requires social science and communication researchers to employ methods able to perform visual content analysis at a large scale. Ongoing advances in machine vision and the ability to automatically detect objects, concepts and features in images provide a promising opportunity to address this challenge, yet it is often not feasible for social science researchers to develop their own custom classifier given the volume of images, resources and technical expertise needed. We therefore propose a research protocol with which existing pre-trained (commercial) models can be used for theory-building purposes despite their black box approach.

We present INCA (short for INfrastructure for Content Analysis), a Python module for collecting, storing, processing, and analyzing a wide variety of media content, including but not limited to news, political debates, social media, forums, and customer reviews. Using Elasticsearch as a database backend and Celery for task management, it makes automated content analysis scalable. INCA's main objective is to enable and promote an integrated workflow. INCA focuses on re-usability of data, processors, and analyses; making all steps of automated content analysis (ACA) accessible to social scientists, without requiring advanced programming skills. Here, we present the aim, implementation and recommended workflow for INCA.