A pilot study at Kennesaw State University using think-aloud protocols suggests undergraduate writers are not simply outsourcing composition to AI tools, but instead make active decisions during the writing process. The research differentiates between what students do cognitively as they compose and what they may report after the fact, aiming to observe how decisions unfold when generative AI is present. The study is positioned against debates where evidence is often drawn from analyses of completed papers or survey-based self-reporting, approaches that may obscure key interaction dynamics. For higher education policy and academic integrity discussions, the implication is that AI use can function as support rather than full delegation—though the piece also notes ongoing concerns in the literature around overreliance, cheating, and potential impacts on critical thinking and engagement.