ChatETP: An AI-Powered Learning Assistant for modern methods of data analysis

With ChatETP, the ETP at KIT is currently piloting a new digital learning assistant designed to support students in physics with data analysis and programming, rather than simply providing answers. The pilot project was developed by the PhD students Johannes Hornung and Jonas Eppelt at the ETP under the guidance of newly appointed professor Jan Kieseler and in close cooperation with GridKa and the Center of Technology-Enhanced Learning (ZML) of KIT. ChatETP comes in two flavours: one is based on the open-source language model llama3.3 hosted on resources of the TOpAS cluster at GridKA. The other flavour is provided through ZML, integrated into Ilias to access lecture materials, and utilises OpenAI models.

At the heart of the tool is a pedagogical approach: rather than providing complete solutions, ChatETP helps students develop their problem-solving strategies. It can assist with debugging, point to relevant resources, and offer feedback on intermediate steps — all to foster independent thinking. Its information base is regularly updated with the topics of the last lectures and exercises.

ChatETP is currently being tested in the course "Modern Methods of Data Analysis". Through this project, the ETP aims to explore how AI-powered tools can be meaningfully integrated into university teaching, not as a replacement for instructors, but as a valuable support for individual learning processes.