Artificial intelligence (AI) is a key technology nowadays and of enormous relevance for our society. A central characteristic of applications and methods from the field of AI is that they are highly interdisciplinary. Both mathematics, computer science, and other STEM as well as non-STEM disciplines are crucial depending on the application at hand. In addition, ethical considerations play an important role. Thus, together with the increasing demand to bring AI education into schools, comes the opportunity to design and implement interdisciplinary projects. Using the example of human activity recognition on smartphones I try to show how interdisciplinary AI projects can be designed focusing on the mathematical aspects while incorporating knowledge from computer science and physics.