Beyond Ease of Use: Dynamics of Technology Adoption and Cognitive Load in AI-Assisted Programming for Non-Technical Students

The integration of Generative AI (GenAI) into non-STEM education introduces cognitive demands that challenge static interpretations of technology acceptance. Evidence on GenAI acceptance is drawn largely from cross-sectional designs applied to general-purpose academic tasks and for computing students who already hold prior technical knowledge, leaving unresolved how acceptance perceptions evolve when non-technical students confront sustained high-load programming work. To address this gap, a pre-test/post-test randomized field experiment with repeated measures was conducted with 117 Chilean Accounting and Auditing undergraduates assigned to a control group (instructional videos; n = 57) or an experimental group (Google Gemini 1.5; n = 60). Participants completed Java programming exercises that increased in cognitive complexity, ranging from basic conditionals to vector operations. The study did not assess objective learning outcomes; instead, it tracked changes in user perceptions through a hybrid motivational framework integrating Perceived Ease of Use and Behavioral Intention with Perceived Enjoyment and Value/Usefulness. Baseline results showed lower initial Perceived Ease of Use in the GenAI group, consistent with initial cognitive friction. Post-intervention analyses indicated that the control group showed a significant increase only in Behavioral Intention, whereas the experimental group showed statistically significant within-group gains across all measured dimensions. These findings suggest that, for non-STEM students, the perceived ease and value of GenAI are not necessarily immediate but may develop through sustained interaction with cognitively demanding tasks. The study contributes to technology acceptance research by framing ease of use as an acquired proficiency in high-friction GenAI learning environments.