8. Conclusions
This study makes several contributions to existing theory. To begin, this is the first empirical academic study to examine user acceptance of IPAs with consideration of the social characteristics attributed to IPAs. Previous studies addressed privacy issues of IPAs (Easwara Moorthy and Vu, 2015), or technical architecture (Chen et al., 2016, Dernoncourt et al., 2017). Mindmeld (2016) reported IPA usage patterns such as usage time, frequency of use, purpose of use, and satisfaction; however, this work was limited to a descriptive survey report. We expect the social aspect of IPA will be highlighted in future studies because IPAs will resemble humans more, as AI technology advances. Second, to our knowledge, no research has been conducted to test the role of PSR in the context of IPAs. This study demonstrated that PSR plays an important role in post-adoption satisfaction and continued usage of IPAs, Therefore, PSR is a powerful theory for anticipating the behavioral intentions of users in the context of human-intelligent computer interaction. Third, this study verified the robustness of the proposed model by introducing new antecedents reflecting risk-related attributes, which has not been investigated in prior PSR research. The empirical results showed that the extended research model had good explanatory power, with an R 2 value of 58.9% for satisfaction and an R2 value of 57.9% for continuance intention. This implies that this new research model creates a useful framework and theoretical basis to explain IPAs, and shows that the application of traditional theories is appropriate to reflect the attributes of this new technology. Yang and Lee (2017) argued that it is necessary to select a base theory carefully and extend the theory to fit the research context, as most technology acceptance theories have limitations.
