Termine
Kalenderdatei
(Für ältere Kalender klicken Sie bitte hier)
Let your students practise high-stakes conversations safely with ExamSim AI avatars [DiL]
21.04.2026, 13:00 - 16:30 Uhr
Hybrid-synchrones Seminar: Center for University Teaching (ZHL) (Nürnbergerstr. 38, 95448 Bayreuth) and online via ZOOM
Oral exams can be particularly challenging for students. They demand subject knowledge, clear communication, spontaneous reasoning, and confidence under pressure - often with limited opportunities for structured practice and targeted feedback.
This seminar introduces ExamSim, an AI-based simulator developed at the University of Bayreuth. It enables students to practise realistic oral exam conversations in a safe environment and provides data-informed feedback on both content and delivery. Together, we will explore didactic potentials, practical use cases, and critical questions around AI-supported training formats. A key focus will be on prompting: participants will learn how to shape exam scenarios, roles, and assessment criteria by writing prompts tailored to their own discipline.
The system combines several components
- A digital avatar acting as the examiner
- A large language model (LLM) generating adaptive and authentic exam dialogue
- Rhetoric and performance analysis (NOVA) tracking selected aspects of delivery (e.g., gaze direction, gestures, speech patterns)
- AI-based content analysis supporting structured evaluation of answers (subject quality and communication quality)
- Immediate, detailed feedback that can be used for student development and self-reflection
ExamSim is currently being used in pilot settings and is continuously refined. Seminar participants will have the option to test the system and explore concrete ways to implement it with their students.
What you will work on in this seminar
- Reviewing your current oral exam formats and identifying where AI-supported simulation can add value
- Developing prompts to create subject-specific exam situations, roles, and levels of difficulty
- Designing assessment criteria and feedback logic aligned with your learning outcomes
- Creating supporting materials to help students practise effectively with the simulator
- Building an implementation plan for the upcoming semester
- Interpreting feedback data for formative student support
Learning outcomes
By the end of the seminar, you will be able to:
- Critically reflect on the opportunities and limits of AI-supported oral exam training
- Assess ethical, legal, and didactic implications of behaviour-tracking and feedback systems
- Design discipline-specific exam scenarios and integrate them into your teaching practice
- Develop a concrete concept for using ExamSim in your courses and iterating it further
More about ExamSim:
https://www.zhl.uni-bayreuth.de/de/projekte/examsim/index.html
Your guide:
Paul Dölle from the Centre for Higher Education Teaching (ZHL) at the University of Bayreuth has been developing digital teaching tools since 2014. He initiated the ExamSim project and brings deep expertise in AI-supported learning, interactive formats, and practical didactics.
More: https://www.zhl.uni-bayreuth.de/de/ueberuns/paul-doelle/index.php
The seminar will now take place as a hybrid seminar. Lecturers from Bayreuth are cordially invited to come to the face-to-face meeting to actively try out the exam simulation themselves.