User acceptance of personal intelligent agent: an extended UTAUT2 with anthropomorphism

Abstract

In artificial intelligence ecosystems, Personal Intelligent Agents (PIAs)—such as Samsung’s Bixby, Apple’s Siri, and Amazon’s Alexa—perform a significant role in communicating with users. PIAs understand the users’ requests and help them as an assistant or a friend through implementing the tasks. Although anthropomorphism can have a substantial influence on users in this process, the current research of PIAs is extremely limited on this concept. Therefore, this study aims to apply the perspective of anthropomorphism to reveal how it affects the users’ perception. In particular, we also examined the relationship between anthropomorphism and the risk factors arising from the use of PIAs. This research employed a new research model that combines UTAUT2 and perceived risk, and further incorporated anthropomorphism as an antecedent. For analysis, structural equation modeling was used based on 170 questionnaire responses. Consequently, the empirical results confirmed that anthropomorphism has a positive impact on users’ perception such as usefulness, easiness, and enjoyment of PIAs. The results also provided evidence that anthropomorphism is able to reduce the perceived risk of users. Furthermore, we identified the factors that influence users’ behavioral intention in the context of PIAs. As AI-enabled technology continues to proliferate, these findings extend the applicability of technology acceptance theories and provide practical implications to service providers.