WISE Agent: AI essay feedback boosts critical thinking in targeted ways

Teachers often struggle to provide timely, customized writing feedback, especially when it comes to fostering critical thinking. To address this, researchers Qi Zhu, Xiaoming Zhai, Yan Zou, and Chunlei Gao built WISE Agent (Writing Improvement and Smart Evaluation Agent), an AI tool designed to identify textual logic gaps and perspective biases in student essays. The system was tested in a three-month intervention with 260 Chinese sixth-grade students, each completing seven themed essays and receiving targeted WISE Agent feedback shortly after submission. Assessments used a rubric adapted from the California Critical Thinking Disposition Inventory (CCTDI), covering seven dimensions like cognitive maturity and open-mindedness.

Surprisingly, the overall improvement in critical thinking scores was non-significant. But deeper granular analysis told a more nuanced story: developmental trajectories diverged by student level. Lower-performing students showed significant gains in evidence selection, while higher-performing students excelled in perspective diversification. This suggests WISE Agent doesn't lift all scores equally—it scaffolds specific cognitive skills depending on where each student starts. The authors argue the tool acts as an effective cognitive scaffold, enabling incremental critical thinking growth through personalized feedback loops. For educators, the takeaway is that AI feedback can support differentiated instruction, but outcomes depend on how the tool is deployed and which skills are targeted.