In this article, I introduce a collaborative annotation activity that supports students in critically examining AI-generated writing in relation to criteria including specificity and complexity. I engage students in collaboratively annotating the AI-generated essays, guiding students to identify instances in which the essays could be more specific--clearly defined or identified--and complex--considering various perspectives or aspects of an issue. I then immerse students in reflecting on how the essays could be revised to be more specific and complex. I posit that this exercise sharpens students' metalinguistic awareness of AI writing. By attending closely to AI-generated essays, students critically interrogate AI writing and invite the construction of more nuanced, complicated ideas. Moreover, this exercise enhances students' agency, recasting the learning process as a collaborative, dynamic endeavor that fosters dialogic interactions between humans and AI writers. Ultimately, I open pathways toward more accessible approaches to supporting students' critical engagements with AI.