ChatGPT and Take-Home Exams: Innovative Approaches to the Responsible Integration of Generative AI in Higher Education
Teaching students to use ChatGPT—AI-driven take-home exams that enhance learning and ensure integrity!
Duration: January 2025 – January 2026
Status: Completed
Educational Level: Tertiary Level
Topic: Artificial Intelligence AI
Keywords: Chatbots, Generative AI, Take-Home Exams, Academic Integrity, Responsible AI
Initial Situation
The project explores how generative AI, particularly ChatGPT, can be responsibly integrated into university assessments. It addresses the challenge of maintaining academic integrity while harnessing the pedagogical potential of AI-based tools. The aim is to design, test, and evaluate new take-home exam formats that promote fair, transparent, and learning-oriented assessment practices. Through literature review, development of two AI-supported assessment methods, and pilot testing in real teaching settings, the project examines how ChatGPT can enhance learning processes without compromising ethical standards. Qualitative and quantitative evaluations measure effectiveness and acceptance among students and instructors. Practical transfer is ensured through workshops and implementation guides that help educators adopt and adapt the new approaches. Ultimately, the project seeks to establish sustainable AI-informed assessment models and contribute to the digital transformation of higher education. By combining scientific rigor with pedagogical innovation, it fosters critical and reflective use of chatbots in academic contexts and strengthens collaboration among BeLEARN partner institutions.
Objectives
The project “ChatGPT and Take-Home Exams: Innovative Approaches to the Responsible Integration of Generative AI in Higher Education” aims to develop new examination formats that enable the didactically meaningful and ethically responsible use of generative AI. It seeks to ensure the integrity and fairness of take-home exams while providing educators with practical tools and guidelines for designing AI-supported assessments. In the long term, the project contributes to improving the quality and sustainability of digital assessment practices in higher education.
Method
The project uses a multi-phase research design combining literature review, development, pilot testing, and evaluation. Based on existing studies, two innovative assessment methods integrating ChatGPT are developed and tested in real university courses. Their feasibility, effectiveness, and impact on learning are examined through qualitative and quantitative data collection. In parallel, practice-oriented workshops train lecturers to apply and adapt the new formats. An iterative approach ensures continuous improvement and practical relevance of the results.
Results
Based on a targeted literature review, we derived an overarching approach for the responsible use of generative AI in take-home assessments. Two core outcomes followed from this work: (1) the development of an ILIAS self-learning course on academic writing with GenAI (e.g., prompting, structuring, quality assurance), which embeds continuous documentation of the type and extent of AI use and a personal prompt portfolio as part of the assessed work; and (2) the design and piloting of an “AI for Students” day as a low-threshold format to build foundational skills. The self-learning course was integrated into two different modules (Master/Bachelor; different formats), piloted, and evaluated using pre-post self-assessments and oral feedback, indicating the expected increase in competencies; minor technical issues were identified and resolved. The “AI for Students” day was implemented with around 120 participants, received positive feedback, and will be continued. In addition, a professional development workshop for teaching staff has been designed, with the first delivery planned for Q3 2026.
Implemented Translation
The project results were successfully translated into educational practice. An ILIAS-based self-learning course on the reflective use of generative AI in academic writing was integrated into two higher-education courses, piloted, and evaluated. In addition, an optional “AI for Students” day was piloted with around 120 participants. Building on these pilots, a higher-education didactics training concept was developed to enable lecturers to use the materials in their own teaching. Implementation partners included the University of Bern, BFH, and FFHS. The first training delivery is planned for Q3 2026.
The project has had a measurable positive impact on educational practice. Pre-post evaluations of the self-learning course and the “AI for Students” format showed expected increases in students’ self-assessed competencies (with cautious interpretation due to varying response numbers). Qualitative feedback also confirmed the practical relevance, clear structure, and usefulness of the materials. Minor technical issues in the self-learning course were identified and have already been fixed. Impact will be further strengthened through the continuation of the student format and the planned lecturer training from Q3 2026.
