Use of AI-based Scaffolding in Writing: Longitudinal Impact on Critical Thinking

• In Event: AERA Roundtable Session Friday 7:45 am - Gold 2
  In Roundtable Session: Scaffolding Learning with AI

Abstract

This study investigates the long-term impact of AI-based scaffolding on critical thinking (CT) in academic writing. Drawing on Vygotsky’s Zone of Proximal Development, the proposed scaffolding framework conceptualizes AI chatbots as “more knowledgeable others” that enhance CT engagement through interactive feedback and reasoning prompts. A three-week pre-post experiment was conducted with 162 undergraduates randomly assigned to three conditions: AI scaffolding with ChatGPT-4o (AIS), ChatGPT-4o only (NAIS), and no AI (NAI). Results from linear mixed-effects modeling revealed that the AIS group significantly outperformed both NAIS and NAI groups in CT development over time (ps <.05). These findings demonstrate the potential efficacy of AI-based scaffolding in fostering critical thinking and highlight its promise for supporting cognitive development in higher education writing contexts.

Authors

• Qi Sun, University at Albany - SUNY

• Alanda Joseph, University at Albany - SUNY

• Cong Wu, University at Albany - SUNY

• Gabriel Schlomer, University at Albany - SUNY

• Tianlin Wang, University at Albany - SUNY