Teaching and Learning with AI Conference

As generative AI becomes increasingly embedded in creative and academic workflows, students must learn not only how to use AI tools—but how to interpret and interrogate them. This session presents a classroom–library collaboration that reframes AI image analysis as a form of pedagogical feedback, positioning machine interpretation as a new rhetorical “audience” in multimodal composition. At Goldey-Beacom College, students in Critical Writing I create visual arguments responding to the common reading Advocate by Eddie Ahn, using either traditional artistic methods or AI image generators. After composing written interpretations of their work, images are analyzed using JSTOR’s Seeklight, an AI tool that generates descriptive readings of visual content. Students then compare the AI’s interpretation with their original intent, examining where meaning aligns, diverges, or is misread. Rather than treating AI output as evaluative or authoritative, this assignment reframes machine description as reverse evaluation: students analyze what the AI “sees,” uncover implicit rhetorical assumptions, identify ambiguity or bias, and revise their writing to clarify meaning for both human and machine audiences. This create → deconstruct → reflect → revise cycle transforms AI from a shortcut into a site of critical inquiry. The session will demonstrate how AI misreadings can deepen students’ understanding of visual rhetoric, authorship, ethical design, and multimodal clarity—while reinforcing academic integrity by centering reflection, revision, and intentional meaning-making.