AI Beacon: Bidirectional Course Evaluation via AI Assistance to Monitor and Enhance Teaching Quality
From feedback to action: enhancing teaching quality through empowering teachers, students, and institutions with AI-enhanced evaluation tools.
Duration: January 2026 – December 2026
Status: Planned
Educational Level: Tertiary Level
Topic: Artificial Intelligence AI, Data Science for Education, Digital Tools
Keywords: Assessment and Evaluation, Artificial Intelligence, Higher Education, Student Feedback, Teaching Quality
Initial Situation
Teaching evaluation is a critical component of quality assurance in higher education, yet current practices are often one-directional and show limited actionability towards improving teaching. We propose an innovative digital solution—AI Beacon—powered by agentic Artificial Intelligence (AI) to support every stage of the course evaluation cycle. AI Beacon offers targeted guidance to:
- support teachers in evaluation design,
- guide students in providing holistic feedback to teachers,
- assist pedagogical advisors in identifying the key action points for teachers to enhance their teaching, and
- help teachers incorporate the necessary improvements to their practices and materials.
AI Beacon leans on the Digital Competency Training Assessment Model (DTCAM), and its operationalization the Digital Training Companion (DTC), which were previously developed by the EPFL LEARN Center. This one-year design science research project will demonstrate the technical feasibility of AI Beacon and pilot its efficiency in addressing the key implementation challenges in course evaluation across higher education contexts. Through iterative refinement and longitudinal validity assessment, AI Beacon promises to become the end solution for enhancing evaluation literacy, improving instructional quality, and streamlining institutional decision-making, paving the way for next-generation teaching evaluation systems in higher education and beyond.
Objectives
AI Beacon builds directly on teaching evaluation theory, introducing cutting-edge innovations through advances in mLLMs, connections to LMS, and analysis of course content. AI Beacon will automate the design of high-quality assessments, support students in providing constructive feedback via AI, and guide the feedback efficiency by generating actionable suggestions. AI Beacon will facilitate the transition towards a new generation of teaching evaluation in which continuous improvement, scalability, bidirectionality, and actionability are supported by AI and guided by human decision-making.
Method
This project consists of two phases:
- development of AI Beacon, a standalone mLLM-powered Swiss-based webservice integrated into the DTC platform, and
- pilot empirical testing of its effectiveness.
Development involves iterative interdisciplinary co-design ensuring refinement, creation of functional modules (teacher support, student feedback, feedback-to-action), and secure technical integration. Case studies at BFH and SUPSI with over 100 participants will combine qualitative and quantitative data to examine assessment quality, feedback usefulness, and adoption.
Planned Translation
AI Beacon will be a marketable, production-ready application, that will bring direct value to educational practice. It will be built to Swiss standards of data security, sovereignty, reliability and robustness. Iterative testing in diverse higher educational institutions (BFH and SUPSI) ensures a good problem-solution fit and quality compliance. Involving three institutions with different languages of instruction provides coverage of the four main languages used in Swiss higher education. Modular programming, clear documentation, and open communication protocols ensure interoperability and allow seamless integration into third-party educational systems, ensuring long-term impact beyond a single platform. As generative AI reshapes education, traditional evaluation methods struggle to measure genuine student learning. AI Beacon assists educators in designing effective formative assessments and guides students in providing constructive feedback. It then transforms this feedback into actionable insights linked to course materials, thus closing the loop between evaluation and improvement. Securely and modularly developed through collaboration between EPFL (education science) and BFH (AI expertise), tested with teachers and students in various contexts at BFH and SUPSI, as well as validated by pedagogical experts, it is ready for adoption across Switzerland’s higher education sector, which consists of over 300’000 students and 25’000 teachers.
