News

The University of Bayreuth has developed ExamSim, an AI-based simulator for oral exams. A digital avatar, powered by AI, acts as an examiner, generating questions, responding to students' answers, and asking follow-up questions. The system supports various difficulty levels and is adaptable to multiple disciplines. AI-driven analytics evaluate students' performance based on eye contact, gestures, voice modulation (via NOVA), and spoken text (transcription & LLM analysis), providing detailed feedback. The pilot phase is scheduled for the winter semester 2024/25.

Benefits for Professors and Lecturers

1. Enhanced Exam Preparation:

   • Students practice oral exams in a stress-simulated yet safe environment.
   • Improves rhetorical skills and structured responses.

2. Data-Driven Teaching Support:

   • The system gives feedback to common weaknesses (e.g., lack of eye contact, unclear articulation).
     
   • Enables targeted feedback and coaching.
     

3. Flexible and Multi-Disciplinary Use:
   

   • Applicable across disciplines (humanities, sciences, law).
     
   • Customizable exam content and questioning styles.
     

4. Seamless Integration into Courses:
   

   • Can be used in tutorials, exam prep, and skill-building exercises.
     
   • Supports adaptive learning paths based on student progress.
     

5. Minimal Setup Effort for Lecturers:
   

   • Predefined templates for various subjects and scenarios.
     
   • Easy content customization to match specific learning objectives.
     

6. Future Scalability:
   

   • Potential for adoption at other universities with tailored content.
   • Supports ongoing research on AI-driven assessments and exam anxiety reduction.

Technology and Research Insights

• Uses LLM for text evaluation, HeyGen interactive avatars and NOVA for behavioral analysis.
  
• Research focuses on reducing exam anxiety and enhancing nonverbal communication skills.
• Collaborative development with AI and digital learning experts.

ExamSim represents a significant advancement in university teaching, enabling innovative, AI-supported oral exam training while enhancing lecturers’ ability to support students effectively.

Interested in testing ExamSim?

For test accounts and custom AI assistant configurations, please contact Paul Dölle from the Center for University Teaching (ZHL) at the University of Bayreuth.