Adoption of AI-assisted coding tools in programming courses: evidence and guidelines for e-learning in engineering

Abstract

The integration of artificial intelligence (AI)-assisted coding tools into programming education offers opportunities to enhance autonomy and immediate feedback in e-learning environments, but it also introduces risks associated with insufficient verification, instrumental dependence, and academic integrity. The overall objective of this study is to design a pedagogical guide for the integration of AI-assisted coding tools into e-learning/hybrid programming courses, based on empirical evidence and geared toward responsible, verifiable use that is compatible with authentic assessment. A non-experimental, cross-sectional, exploratory-descriptive quantitative study was conducted at the Bolivarian University of Ecuador with a sample of 111 students. An ad hoc questionnaire with 28 items (Likert 1–5) was administered, organized into four dimensions: use and frequency, perceived usefulness, trust/control, and ethics/risks. The instrument showed excellent internal consistency (overall α = 0.9736). The results show high perceived usefulness (M = 3.71) and moderate ethical awareness (M = 3.57), along with heterogeneous adoption (use and frequency: M = 3.29). Significant positive correlations were observed between dimensions and differences between inconclusive degrees in most comparisons. Based on the diagnosis, an operational pedagogical guide for e-learning/hybrid learning was designed (rules of use, traceability log, mandatory verification, and authentic assessment), validated by expert judgment (n = 12, one round) with an overall average Aiken's V of 0.87. It is concluded that effective adoption requires integrated guidelines that articulate productivity, deep learning, and academic integrity.