Extended CAITIZEN: A PLS-SEM Study of Sustainable AI-Assisted Citizenship Innovation

Context. Artificial intelligence is transforming higher education by reshaping learning, creativity, decision-making, and civic participation. This study examines CAITIZEN—Citizenship Assisted by Artificial Intelligence for Sustainable, Ethical, and Networked Formation—as an extended model for validating AI-assisted sustainable citizenship as an innovation for sustainable development, aligned with SDG4 and SDG9. Problem. Although the original CAITIZEN model was qualitatively grounded as an ethical–cognitive–social framework, its explanatory and predictive capacity had not been empirically tested. AI education still prioritizes efficiency, automation, and technical adoption, with limited evidence on how critical AI literacy, ethics, data justice, human–AI collaboration, and metacognitive prompting contribute to sustainable AI-assisted citizenship. Purpose. This study validates the extended CAITIZEN model through PLS-SEM by examining how Critical Artificial Intelligence Literacy (CAIL) enables Ethical Awareness and Responsibility (EAR), Awareness of Fairness and Data Justice (AFDJ), Human–AI Creative Collaboration (HAIC), and Metacognitive Transparency in Prompting Practices (MTPP), and how these capacities predict CAITIZEN. Methodology. This research builds on a previous qualitative phase conducted in Guadalajara, Jalisco, Mexico, during July–December 2025, and complements it with an explanatory-predictive quantitative design using SmartPLS 4.1.1.8 to assess reflective constructs and predictive relevance through PLSpredict. Theoretical and Practical Findings. Results show that CAIL significantly predicts EAR, AFDJ, HAIC, and MTPP, confirming its role as foundational antecedent. AFDJ, HAIC, and MTPP significantly predict CAITIZEN, whereas EAR does not show a direct effect. Predictive relevance is confirmed because all Q²_predict values for CAITIZEN indicators are positive and all PLS-LM RMSE differences favor PLS-SEM. Originality. The study transforms the qualitative CAITIZEN model into an empirically validated explanatory-predictive structure. Conclusions and Limitations. The extended CAITIZEN model provides a measurable framework for responsible AI education and sustainable innovation. Limitations include non-probabilistic sampling, cross-sectional design, and student sample.