Generative AI in Action: Using AI to Scaffold Economic Writing Skill Development

This paper addresses a growing challenge in higher education: the shift from take-home essays to in-person writing exams in response to generative AI. While in-person assessments safeguard academic integrity, they also remove the extended time students once had for deep thinking, drafting, and reflection. To bridge this gap, we introduce a scaffolded pre-exam approach in a second-year Microeconomics Theory course. The intervention combines three steps: timed practice writing, AI-generated feedback guided by an instructor rubric, and structured student reflection. Drawing on self-regulated learning theory and the STRIVE framework for AI-integrated assessment, the exercise aimed to develop economic writing skills while teaching critical evaluation of AI suggestions. Survey data revealed that students found AI feedback helpful for improving argument organization and clarity, though concerns emerged about maintaining authentic voice and the AI's limited grasp of economic reasoning at the time. Building on these insights, we propose the SCAN-X framework (Scaffolded Critical AI-Nested eXams) as a generalizable model for designing scaffolded pathways that link AI-supported pre-exam practice with in-person assessments. The intervention’s success depends on explicit scaffolding of metacognitive reflection and critical AI literacy, offering a model for courses where writing is assessed but not explicitly taught.