<![CDATA[The development of Generative Artificial Intelligence (AI), such as ChatGPT and Gemini, has transformed the way students acquire, verify, and interpret information in the learning process. However, the cognitive and epistemological dynamics of students interacting with AI remain underexplored. This study aims to: (1) analyze how students position AI as a learning assistant or a learning dependency; (2) investigate the verification strategies students use to assess the accuracy of AI outputs; (3) examine the impact of AI use on learning autonomy and Self-Regulated Learning (SRL) abilities; and (4) elucidate students’ ethical and epistemological reflections in assessing truth in the digital era. This research employs a qualitative approach with a descriptive phenomenological design, involving 10 students from the Islamic Education Study Program at UIN Sunan Kalijaga who actively use generative AI in academic activities. Data were collected through in-depth interviews and analyzed using thematic analysis techniques. The findings indicate that students fall along a spectrum of AI utilization, ranging from supportive learning use to cognitive dependency. Verification strategies vary as well, from rigorous literature-based verification to algorithmic trust, relying on the logic or academic appearance of AI responses. AI usage affects SRL differently: integrative use strengthens planning and reflection, whereas substitutive use leads to cognitive offloading, weakening learning autonomy. Furthermore, students’ ethical and epistemological reflections reveal differing levels of understanding regarding the probabilistic nature of knowledge generated by AI.]]>
