Teacher-LLMs Integrated Feedback (T-LIF) system
Background Information
Teachers at TU/e increasingly work with large and diverse student cohorts, making it difficult to provide timely and meaningful feedback. In many courses, feedback is delivered through automated templates or time intensive manual processes. While templates ensure consistency, they often result in generic messages that students find difficult to act upon. Manual feedback, on the other hand, places a heavy burden on teachers and does not scale well.
These challenges are particularly visible in conceptually demanding technical courses and in courses that emphasize professional and personal development. Students need feedback that is specific to their progress, strengths, and learning strategies, but teachers lack the time to tailor messages for every individual. As a result, feedback may arrive too late or lack the depth required to support self-directed learning.
Recent developments in large language models and learning analytics offer new opportunities to address this gap. Earlier pilots at TU/e have shown that AI supported feedback can reduce workload and improve consistency but also highlighted the limitations of fully automated approaches without teacher oversight. This pilot builds on TUe education innovation by integrating AI tools directly into the teacher feedback workflow. By combining learning analytics with LLM supported feedback generation, the project aims to enhance feedback quality while keeping teachers in control of pedagogical decisions.
Aim of the project
The aim of this project is to develop and validate the Teacher LLMs Integrated Feedback (T-LIF) system in two departments (EE and IE&IS). T-LIF is an environment that supports two courses’ teachers in crafting, improving and delivering personalized feedback for over 350 students. The system combines learning analytics with LLM technology to track individual learning paths and transform analytical and pedagogical insights into context-aware and learner-centered feedback.
For students, the project aims to improve learning effectiveness and well-being by providing timely feedback that clarifies strengths, identifies areas for improvement, and suggests concrete next steps. This supports students’ self-regulation and learning success.
For teachers, the project aims to significantly reduce feedback time while improving consistency and depth. Teachers’ decision-making in providing learning suggestions and
feedback will be supported by evidence from learning analytics and by LLMs that enable efficient drafting, formatting, and rephrasing.
At an institutional level, the project aims to establish a responsible model for AI-supported feedback that respects educational values and supports diverse learning paths. By embedding the system into existing digital learning environments, the project contributes to scalable, data informed, and student-centered education across TU/e.
For more information, please contact:
