Skip to Content
Man in gray jumper leaning against a white pillar

What Does Ice Cream Have to Do With AI?

  • For the Region
  • Vision

The goal of the “KI B³” and “MEKI” cooperation projects is to bring Artificial Intelligence (AI) into vocational training. Prof. Stefan Wagner from TUM Campus Heilbronn has developed learning software and a chatbot and now wants to use Large Language Models to provide personalized support for students.

Is Germany lagging behind when it comes to Artificial Intelligence? Stefan Wagner, Professor of Software Engineering at the TUM School of Computation, Information and Technology at Campus Heilbronn, takes a more optimistic view: “We have nothing to be ashamed of. I know a lot of great start-ups and established companies that are starting activities in the field of AI. We are also at the forefront of research.”

To stay ahead in the AI league in the future, qualified young talent is needed. And here is indeed a need to catch up: Wagner does not yet see the topic of AI represented in most training courses. However, he does not consider drastically changing the curricula to be a good idea. He would rather focus on teaching AI topics in further training measures.

 

Trainees in All Professions Are Welcome

 

This is where the now completed project “KI B³ – Bringing Artificial Intelligence into Vocational Training” comes in, in which Wagner represented the TUM as project manager. As part of the project, which was funded by the German Federal Ministry of Education and Research (BMBF) with almost 670,000 euros, the six partners – in addition to the TUM, the Chambers of Commerce and Industry (IHK) Reutlingen, Karlsruhe and Stuttgart Region, the University of Stuttgart and the Ludwig Maximilian University of Munich – have designed an additional qualification (ZQ) and further training courses to become a certified professional specialist or a Bachelor Professional in “Artificial Intelligence and Machine Learning (IHK)”. The target groups for the ZQ are apprentices in all professions, while the two further training courses are aimed more at IT professionals.

The ZQ first teaches the basics of AI, machine learning and data collection. It then moves on to the topic of data analysis: “There's the nice learning game of the ice cream seller, who can calculate how much ice cream they need to prepare based on the weather forecast,” says Wagner. At the end of the ZQ, the course assesses the opportunities and challenges of AI and addresses ethical questions.

 

From Programming to Practical Application

 

The further training to become a certified professional specialist is more technically oriented, teaching participants programming and data analysis with Python, as well as how to train AI models. Finally, the continuing education program for Bachelor Professional focuses on practical application: Participants develop a complete AI system using exercise sheets and project examples – from data collection and training to the transfer of predictions to other systems.

TUM and the University of Stuttgart defined the course content and created the AI-related modules. In addition, Wagner and his team have taken the lead in developing learning software that participants can use to independently deepen their knowledge. Essentially, this consists of plugins for the Moodle learning platform. Also the interactive server application JupyterHub, a website where text can be mixed with programming code, was integrated in Moodle. The plugins have since been published as open source and as official Moodle plugins in the Moodle Plugin directory. The researchers at TUM have also contributed to a chatbot that motivates students and suggests suitable learning content to them.

 

Cooperation as a Model for the Future

 

The follow-up project “Achieve More with AI” (MEKI), with almost identical project partners, started immediately after KI B³. The aim of the three-year project is to prepare the content even more individually in order to optimally support participants at all levels. Large Language Models (LLMs) are to be used at various points in the process. “At TUM, we want to see how LLMs can help students understand software and code,” explains Wagner. ”You can use LLMs to have code explained or generated. We want to explore this interaction in more detail and pursue the idea already contained in KI B³: We use AI to reach people individually.”

Will KI B³ and MEKI change vocational training permanently? Wagner sees two main aspects in favor: “On the one hand, such projects could help us to break up the traditional teacher-centered teaching at vocational schools and to anchor independent learning more firmly in training.” On the other hand, he sees the collaboration between universities and chambers of commerce and industry as a kind of blueprint: “Such cooperation projects offer the opportunity to package the expertise of universities in a form that can be easily applied in vocational training. The IHKs are the ideal partner for this because they have close contact with companies and students. This combination could be a model for the future.”