A pilot study led by researchers using think-aloud protocols at Kennesaw State University provides new evidence about how undergraduate writers interact with generative AI during composition. Rather than relying on completed papers or self-reported surveys, the study captures decision-making as it happens, showing that students are not simply outsourcing authorship to AI in a uniform way. The research approach focuses on process—how students make choices about prompts, revision, and tool output while drafting. The study also sits inside an ongoing higher-ed debate about generative AI’s role in writing instruction, including concerns about overreliance, cheating, and potential degradation of critical thinking and engagement. For colleges and universities, the immediate implication is assessment design: if AI use patterns vary by drafting behavior, institutions may need more nuanced academic integrity policies and classroom guidance that address process-based risks rather than assuming identical outcomes. As regulators and accrediting conversations increasingly turn to student learning evidence, this kind of process-level research can shape how instructors interpret tool-assisted writing and set expectations for responsible use.