Self-Perceived Knowledge, Self-Efficacy, and Reported Integration of Generative Artificial Intelligence Among University Faculty: An Exploratory Profile Analysis

Abstract

Generative artificial intelligence (GAI) adoption in higher education raises questions about the relationships among perceived knowledge, confidence, attitudes, and reported teaching practices. A cross-sectional study of 149 faculty members at an Ecuadorian public university examines four self-reported constructs, with Cronbach’s α ranging from 0.841 to 0.981. An exploratory three-group partition describes higher-scoring (44.3%), intermediate-scoring (39.6%), and lower-scoring (16.1%) respondents, although the silhouette criterion favors two groups (0.383 versus 0.299). Self-efficacy correlates with reported integration (r=0.812; approximate 95% CI [0.749, 0.860]). A complementary three-predictor regression reconstructed from the reported correlation matrix yields R2=0.668 and a self-efficacy coefficient of β=0.758 (approximate 95% CI [0.630, 0.885]). This model excludes demographic and frequency covariates and does not replace the original adjusted analysis. Reconstructed aggregate counts show an association between profile and grouped use frequency (conditional Monte Carlo p=0.0037), whereas the gender association remains inconclusive (p=0.089). Measurement overlap, self-reporting, and unverified cluster stability limit interpretation. Normalizing response ranges does not establish that faculty practice more than they know. The results support evaluating faculty development that combines conceptual and critical AI literacy with GAI-specific self-efficacy; causal effectiveness remains untested.