Abstract: This study examines the Competence Paradox in AIassisted learning. The students need digital and ethical competencies to understand how to use these AI tools, but they again use those same AI tools to develop such competencies. Using a sequential dominant mixed-methods approach, the research investigated how IT students engage with generative AI platforms such as ChatGPT, Copilot, Gemini, and Claude. Quantitative findings from 200 participants uncover strong correlations between AI literacy and awareness of ethics, instructional support, and perceptions of AI's alignment with SDG4: Quality Education. Qualitative insights emphasize recurring tensions: confidence built through informal learning, uncertainty about ethical boundaries, inconsistent guidance provided by educators, and dependence on AI that may undermine deeper learning. The results confirm hypotheses that, though students are digitally capable, their responsible and equitable use of AI is wholly dependent on structured support and clear ethical frameworks. In addressing the Competence Paradox, this study contributes to the discourse on sustainable innovation in education and provides actionable implications for curriculum design and policy development.