SPARK: AI-powered Reflective Writing
AI-powered Reflective Writing: Developing an Open-Source Tutor to Enhance Students’ and Educators' Metacognitive Skills
Factsheet
- Schools involved Business School
- Institute(s) Institute for Digital Technology Management
- Research unit(s) Human AI Learning Systems
- Strategic thematic field Thematic field "Humane Digital Transformation"
- Funding organisation SNSF
- Duration (planned) 01.06.2026 - 30.06.2027
- Head of project Prof. Dr. Thiemo Wambsganss
- Project staff Prof. Dr. Thiemo Wambsganss
- Keywords Open Source Learning Tool, AI-based Feedback, Metacognition, Reflective Writing
Situation
Metacognition and reflective writing is a critical skill for students, fostering deep learning, self-awareness, and hence, crucial to achieving the United Nations' Sustainable Development Goals. However, many learners struggle with structuring their reflections effectively and receiving timely, personalized feedback. While Artificial Intelligence (AI)-powered writing support tools have shown promise in improving academic writing, their role in fostering reflective writing remains largely unexplored. As universities seek innovative ways to support students’ metacognitive development, understanding the potential and limitations of AI-powered reflection tools is both timely and necessary.
Course of action
This 12-month project is designed to investigate whether and how conversational tutoring can support students in their reflective writing skills. Specifically, the focus lies on the examination of two AI-interaction modes—(1) structured guidance for writing reflective texts and (2) offering adaptive feedback—and whether these modes differentially affect the quality of the final reflection texts.
Result
Further, we will investigate the effect of our reflective guidance on students' perception of their reflection skills and their progress. Offering guidance and feedback should help students gain confidence in their skills and allow them to experience self-efficacy in regard to their ability to write reflective texts. However, while AI-powered writing aids have shown promise in some educational domains, their application to sustained reflective practice is nascent and unproven. The project’s success is uncertain—negative findings are equally valuable as they provide novel insights into how AI can (or cannot) support deeper reflection skills.
Looking ahead
The project’s innovative character lies in its combination of adaptive AI-powered feedback with a structured reflection framework, providing a human-centered approach to reflective writing support. By analyzing both qualitative and quantitative indicators of student engagement and skill development, this project can advance the understanding of how AI can support reflective learning processes. Findings will be disseminated to educators, along with practical recommendations for integrating the tutor as an open source tool in educational institutions in Switzerland. This research contributes to a growing conversation on responsible AI in education, offering actionable insights into the design, implementation, and impact of AI-powered reflection tools.
