Editorial: Conversational AI

Stephan Raaijmakers, Anita H. M. Cremers, Emiel Krahmer, Matthijs Westera

Abstract: Given these significant developments, one could be tempted to think that Conversational AI has come of age. Yet, a full slate of unresolved problems and research questions remains. Crucial debates surround the societal impact of large language models and the future of NLP, the environmental impact of training regimes as well as mass adoption, the impact and prevention of bias, and possible copyright infringement of training data. Central research topics in the field of Conversational AI are to a large extent orthogonal to the underlying technology, including large language models. This special issue of Frontiers addresses a number of such topics: the human perception of conversational agents and the effects of social cues exhibited by conversational agents on humans, the role of information presentation in hybrid conversational systems, the usage of carefully annotated data in addition to raw textual observational data, and the emergence of communicative patterns between humans and machines.The paper by Peter Blomsma, Gabriel Skantze and Marc Swerts addresses the perception by human interlocutors of personality traits displayed by embodied conversational agents. These authors demonstrate through a comparison of human-human and human-AI interaction that dynamic social feedback cues, in particular head nodding, correlate with human-perceived personality traits. With Conversational AI becoming increasing multimodal and embedded, these findings will be of practical interest to...