Mission Short

Automate correction of assignments and exams in higher education

The Challenge

The amount of time teachers spend correcting assignments. This is often repetitive and takes up too much of a teacher's time, distracting them from productive tasks. These engagements can often take up to 10 hours per week. This affects both students and teachers. Teachers have less time to work on research and teaching, while students receive sub-optimal feedback, if any.

The solution

A locally hosted AI Agent is informed on the course material and grading expectations (RAG). The LLM then corrects the assignments and provides detailed feedback to each student. The AI Agent will be built to have a human-in-the-loop system so that corrections can be challenged, and the final result is controlled by the educator. Local hosting is important to ensure compliance with GDPR. A model fine-tuned for the specific subject provides the highest accuracy (STEM subjects, programming, history, ..) The model can also be trained on past data of corrections.