Overview

A university professor at École Polytechnique de Louvain conducted an experimental exam allowing students to use chatbots with full transparency requirements. Only 3 out of 60 students actually chose to use AI assistance despite being explicitly permitted to do so.

The Breakdown

  • Open-book, open-chatbot exam format - students could use AI tools but had to declare usage in advance and share all prompts
  • Full accountability requirement - students took complete responsibility for any AI-generated mistakes in their answers
  • Unexpectedly low adoption rate - only 5% of students used chatbots despite explicit permission and no penalties
  • Student motivation survey - professor gathered data from half the class to understand why most avoided AI assistance