Inteligencia artificial y competencias de aprendizaje en programación

The expansion of artificial intelligence (AI) in higher education programming raises questions about its influence on learning autonomy and critical thinking. This stems from theoretical concerns about technological dependency and from the limited regional empirical evidence on how these effects vary by experience level. This quantitative study examines whether experience level (novices vs. advanced) and AI usage frequency associate with these constructs among computer science students in Costa Rica. Using a cross-sectional design, 99 students from Universidad Hispanoamericana de Costa Rica were classified as novices (n = 79) or advanced (n = 20), via the Dreyfus & Dreyfus (1986) model with 5 weighted indicators. Accordingly, autonomy (а = 0,685) and critical thinking (а = 0,742) were measured, and both constructs were analyzed using Welch t-tests, Pearson correlations, and multiple regression. Autonomy showed suggestive evidence of being higher in advanced students [t(49,2) = 2,079, p = 0,043, g = 0,386], though nonparametric analysis did not confirm this [U = 614,5, p = 0,125]. Critical thinking showed no significant difference [t(42,1) = 1,816, p = 0,077, g = 0,361], and AI usage frequency was not associated with autonomy [r = 0,166, p = 0,101] or critical thinking [r = 0,123, p = 0,226]. Both constructs showed robust correlation [r = 0,552, p < 0,001; р = 0,533, p < 0,001]. The 88,9 % active modification rate suggests students use AI as cognitive support, potentially explaining the null associations. Given that 85,9 % support regulated use and 10,1 % favor prohibition, these preliminary findings suggest that promoting ethical regulated AI use may be preferable to restriction.