Understanding perceived learning and the role of digital self-efficacy in generative AI adoption among STEM and non-STEM students

This study examined how Digital Self-Efficacy (DSE) shapes the adoption of generative artificial intelligence (AI) tools and their relationship with students’ Perceived Learning in higher education. STEM refers to science, technology, engineering, and mathematics, while non-STEM includes disciplines such as the social sciences, humanities, and arts. Focusing on widely accessible generative AI applications for writing, coding, and information processing, the study integrates the Technology Acceptance Model (TAM) and Social Cognitive Theory (SCT). Survey data from 677 undergraduates in Ecuador were analyzed using Partial Least Squares Structural Equation Modeling and complemented with follow-up interviews. Results indicate that higher DSE increases perceived usefulness and ease of use of AI tools, which in turn foster positive attitudes and stronger intentions to adopt them. Behavioral intention to use AI is positively associated with perceived learning, with this effect being stronger among no-STEM students. Qualitative findings further show that STEM students primarily use AI fortechnical or task-specific work, whereas non-STEM students reported broader gains in digital confidence and efficiency. Overall, the study highlights the central role of digital self-efficacy in enabling meaningful adoption of generative AI in higher education, while acknowledging that findings are based on self-reported perceptions from a digitally active student population.