2026 ASEE Annual Conference & Exposition

The incorporation of artificial intelligence (AI) into higher education not only introduces tools that automate tasks and support problem-solving; but it also reshapes teaching and learning approaches. In engineering—and particularly within the computing curriculum—this transformation is critical because professional preparation demands both technics competencies and critical–reflective skills. Courses such as programming and databases therefore provide a strategic setting: students work with formal languages, logical structures, and complex problems, an ideal context to examine how AI literacy—understood as the ability to understand, apply, and critically evaluate these technologies—relates to metacognitive awareness, defined as the capacity to plan, monitor, and regulate one’s own learning. This articulation is crucial for promoting autonomous learning and fostering an ethical, critical engagement with technology. This study examines that relationship in undergraduate students enrolled in programming and database courses using a two-block questionnaire: the Meta AI Literacy Scale (MAILS), which assesses conceptual knowledge, use/application, ethics, self-efficacy, socioemotional self-competence, and notions of AI creation; and the Metacognitive Awareness in AI Use (MAI-U), an adaptation of classic metacognition frameworks contextualized to AI use and integrating technics components (system limitations and trustworthiness) and ethical components (bias, transparency, and responsibility). Using a cross-sectional design and quantitative approach, the instrument was administered in class, with informed consent, to a sample of 129 students at a private engineering school in Chile. Subscale scores and composite indices (e.g., global MAI) were constructed, reliability and descriptive statistics were estimated, and associations were analyzed via correlations and linear regressions with robust standard errors and collinearity diagnostics, modeling as outcomes AI use and application and AI creation, and including as predictors global MAI, AI self-efficacy, and AI self-competence. Preliminary results indicate high levels of self-efficacy and conceptual AI literacy, as well as comparatively lower levels in creation and detection. The observed associations show a robust convergence between AI metacognition (MAI) and the various literacy facets (knowing, using, detecting, and ethics). Provisional regression models suggest that MAI is the primary predictor of AI use and application as well as conceptual AI knowledge, with self-efficacy providing an additional contribution; perceived competence shows no unique effect once the former are controlled. Taken together, these findings enable concrete curricular actions in computing education: strengthening metacognition applied to AI in early courses to boost use and application; complementing with strategies that reinforce self-efficacy; and addressing gaps in creation and detection through authentic tasks and clear quality criteria. This alignment enables programs to connect learning outcomes, assessments, and instructional support, allowing them to monitor cohort progress and make informed decisions based on evidence. Ultimately, the results support a competent, ethical, and sustainable adoption of AI in the education of future engineers.