Abstract This study examines how ChatGPT influences high school students’ cognitive and behavioral engagement in Physics learning, drawing on survey data from 251 students at Vo Minh Duc High School in Ho Chi Minh City. Although generative AI is increasingly adopted in education, little is known about how Vietnamese students use ChatGPT in a cognitively demanding subject like Physics-representing a notable research gapin a cognitively demanding subject like Physics, which represents a notable research gap. Using a mixed-methods design, the study explores how students employ ChatGPT, how they perceive its usefulness and reliability, and how its use shapes learning behaviors. Findings indicate that students commonly use ChatGPT as a self-study aid and perceive it as helpful for understanding complex concepts, yet many express concerns about accuracy and show signs of overreliance when using the tool without guidance. The study highlights the dual impact of ChatGPT-supporting learning while potentially reducing independent thinking-and contributes evidence potentially reducing independent thinking and contributes evidence to inform AI literacy development, teacher training, and pedagogical frameworks for responsible integration of generative AI in secondary education.

This study examines the relationships between Training Quality (TQ), Institutional Support (IS), Faculty Readiness (FR), and Perceived Effectiveness (PE) in the context of artificial intelligence (AI) integration in higher education institutions (HEIs) in Vietnam. Employing Partial Least Squares Structural Equation Modeling (PLS-SEM), data from 418 faculty members from higher education institutions (HEIs) in Vietnam were analyzed to identify key factors influencing the adoption of AI in teaching. The findings reveal that both TQ and IS significantly enhance FR, underscoring the critical importance of comprehensive training programs and institutional resources for preparing faculty to adopt AI. Furthermore, FR has a substantial impact on PE and serves as a mediator between TQ and PE, as well as IS and PE. This highlights the pivotal role of faculty readiness in transforming training and support into perceived improvements in teaching effectiveness. The model demonstrates high predictive relevance for both FR (Q² = 0.55) and PE (Q² = 0.60), suggesting the robustness of the theoretical framework. Despite the study’s limitations, including its focus on Vietnamese HEIs and cross-sectional design, it provides valuable insights for designing effective faculty development and institutional support strategies to facilitate AI integration