Generative AI tools like ChatGPT and Copilot are rapidly reshaping how students approach computing tasks - but what do instructors actually see when these tools enter the classroom? This study explores how computer science faculty perceive the impact of generative AI on students’ reasoning and problem-solving processes. Through semi-structured interviews with six experienced instructors from four different institutional contexts, the study investigates how educators interpret students’ cognitive engagement, assess learning, and adapt their pedagogy in light of AI use. Early analysis suggests a mix of apprehension, creativity, and philosophical tension. Some participants described new reflective strategies - such as requiring students to document prompts and credit AI contributions - to make thinking more visible. Others voiced unease about students’ growing emotional distance from their work, their uncritical trust in AI’s authority, and the institutional impulse to impose one-size-fits-all policies. Across interviews, a shared theme emerged: instructors are grappling with how to see inside what one called the “cognitive black box” of student AI use. By surfacing these perspectives, this project aims to move the conversation beyond fear or blind adoption toward curiosity and constructive pedagogy. The findings offer a foundation for designing supports and interventions that help faculty engage critically - and empathetically - with AI in computer science education.