The Effect of AI Engagement on Generative AI Adoption

This project investigates the influence of user engagement on the adoption and sustained utilization of Artificial Intelligence (AI) technologies. It aims to fill the research gap in understanding how different facets of engagement—cognitive, emotional, and behavioral—affect the adoption of AI across various domains. The central research question is how users' engagement with AI influences their adoption intentions. The study employs Social Cognitive Theory (SCT) as its theoretical lens.

Background and Hypotheses: The study defines engagement as a state of being occupied with an activity that displays energy, involvement, and efficacy. It introduces AI engagement as a second-order construct comprising affective, behavioral, and cognitive dimensions. Social Cognitive Theory is used to explain the relationship between AI engagement and its antecedents and consequences. Two hypotheses are proposed:

• H1: AI artifact familiarity is positively related to AI Engagement.
• H2: AI Engagement is positively related to intention to use the AI artifact.

Method: A sample of 258 undergraduate students interacted with a generative AI tool, DALL.E2, and then completed surveys on their familiarity with the AI tool, AI engagement, and intention to use the AI tool.

Results: Structural equation modeling was used for analysis. Both hypotheses were supported, indicating a positive and significant relationship between AI tool familiarity and AI engagement, and between AI engagement and intention to use the AI tool.

Discussion and Conclusions: The study investigates the role of user engagement in the adoption of Artificial Intelligence (AI) tools, employing Social Cognitive Theory as a theoretical framework. The results indicate a significant relationship between familiarity with AI tools and user engagement, as well as between engagement and the intention to use such tools in the future. The research contributes to Information Systems theory by introducing AI engagement as a second-order superordinate construct that predicts intention to use. Practically, the findings suggest that organizations should adopt deliberate strategies to familiarize employees with new AI tools to enhance self-efficacy and engagement. However, the study acknowledges limitations such as the low reliability of the AI familiarity scale and potential sample bias. Future research aims to replicate the study with different AI forms and additional moderating constructs like AI anxiety and social presence.