Acceptance of Generative AI Tools: An Extended UTAUT and SEM-ANN Hybrid Model Approach with Chinese University Students

University students' adoption of generative AI tools in educational contexts is investigated in this research through two complementary studies. A Jordanian-based cross-sectional study (N=95) initially established the middle students' familiarity (M=3.14, SD=0.81) with generative AI writing tools, along with significant concerns about misinformation/data security (M=3.35, SD=0.85), but high awareness regarding advantages for creativity simulation and innovation facilitation (M=3.62, SD=0.81). Second, a Chinese UTAUT-based research (N=361) employing SEM-ANN hybrid analysis found that DeepSeek adoption intention is significantly determined by Perceived Price (β=0.346), Performance Expectancy (β=0.244), Effort Expectancy (β=0.233), eXplainable AI (β=0.126), and Facilitating Conditions (β=0.084), while Social Influence did not appear to be significant (β=-0.019). Moderator analysis found gender and education level to significantly moderate some predictor-intention relationships. Synthetically, these findings demonstrate: (a) cultural consensus in recognition of AI's value for learning in various capacities despite technical challenges, (b) the critical role of cost disclosure, usability, and explainability in adoption, and (c) stratified intervention needs to address technical instruction, ethical literacy, cost reduction, and maximization of usability - particularly through personalized learning deployments. The study supports the UTAUT's efficacy for use in AI adoption studies while enriching methodology integration through SEM-ANN methods.