Samenvatting

Rapid progression in generative artificial intelligence technology facilitates faster and easier conversational agent design. However, deploying conversational agents is challenging and can easily have adverse effects. To this end, more conceptual clarity on how to design for, for example, trust, in conversational agents is needed. Ideally, using theories could provide conceptual clarity and guide and ground conversational agents designed for trust. We conducted a scoping review on how systematically trust is being designed for in conversational agents. The review shows that although the conversational agent work is well grounded, the variation in referential theories used, manipulations chosen, and measures used is high. Moreover, the relation between these dimensions is not always explicitly explained. This hinders the abstraction of more generalizable research findings. We conclude with the suggestion for the development of more intermediate levels of knowledge in the context of designing for trust in conversational agents.