To path indicator Subpages of “Research“ To navigation To quick access To footer with other services
To content
Department of Statistics
Studenten im Sommer auf dem Campus Nord, rote Zahnräder im Hintergrund © Roland Baege​/​TU Dortmund

Exploring the impact of AI-generated, adaptive prompts on students’ understanding and learning of programming

Project description

Programming is associated with critical skills such as computational thinking and problem-solving, and has been shown to benefit creative thinking, mathematical skills, and metacognition. Nonetheless, research on misconceptions and failure rates in introductory programming demonstrates that students often struggle with basic programming concepts even after completing the introductory courses successfully. One of the challenges novice programmers face when practicing code-writing is interpreting and addressing compiler and interpreter warnings and error messages. 

This project aims to answer two research questions as a first step to address this challenge:

  • RQ1: What is the optimal prompting condition in order to have an LLM generate hints that best help students understand and learn from compiler errors?
  • RQ2: How can we help students with exhibiting prompting behavior that leads to optimal tailored feedback/support from the LLM model?

Working toward answering these research questions, we will enhance an interactive programming environment (e.g., Jupyter Notebook) with an AI student model that will allow us to monitor and model students’ knowledge state. We rely on compiler error messages to assess students’ progress and use these messages as opportunities for learning. To do so, we  couple the AI student model with a Large Language Model (LLM) to explain compiler error messages in the way a student would use it in their everyday practice.

Principal investigators

  • Prof. Dr. Irene-Angelica Chounta, Department of Human-centered Computing and CognitiveScience, University of Duisburg-Essen
  • Prof. Dr. Paul Bükner, Department of Statistics, TU Dortmund University
  • Prof. Dr. Falk Howar, Department of Computer Science, TU Dortmund University
  • Prof. Dr. Nils Köbis, Research Center for Trustworthy Data Science and Security and University of Duisburg-Essen
  • Prof. Dr. Nikol Rummel, Ruhr-University Bochum and Center of Advanced Internet Studies (CAIS) gGmbH
  • Prof. Dr. Markus Pauly, Research Center for Trustworthy Data Science and Security and Department of Statistics, TU Dortmund University

Funding

The project is funded as an "Incubator project" by the Research Center Trustworthy Data Science and Security, one of four research centers within the UA Ruhr that are funded by the Ministry of Culture and Science of the State of North Rhine-Westphalia.

To top of page