Abstract

This paper proposes a systematic review of the empirical research investigating why artificial intelligence (AI) devices are accepted or rejected. The aim is to discover and examine pivotal determinants related to AI acceptance, to resolve contradictions within the literature and to detect potential research areas that are unexplored, thus promising a holistic understanding of how humans interact with AI technology. The review highlights significant gaps in the literature with regard to how expectations, contextual factors and emotions are associated with AI acceptance. Effort expectancy, social influence, and anxiety are commonly investigated; however, the findings are conflicting. Hedonic motivation and trust are found to be significant antecedents for acceptance, but their mediating effects with other emotional and contextual factors are still less researched. Differences in methodology, in population, and in the AI applications evaluated, may have contributed to conflicting results. Such findings imply that AI acceptance is multidimensional in nature and cannot be comprehended by isolated constructs. Future research needs to focus more on integrated models that incorporate the interplay of expectations, affective responses and situational factors, taking into account cultural and organizational contexts. Working on these dimensions will facilitate the development of AI systems that better serve human needs and ideals. This review adds an important dimension to the literature on AI adoption, drawing together fragmented and, at times, contradictory evidence, highlighting areas in which much remains to be known and setting the agenda for future research.

Statistics