Cross-faculty collaboration on AI literacy¶

Faculty of Informatics¶

Our department has a strong interdisciplinary nature, fostering ties with many research departments at TU Wien, which are based on collaborative research projects and co-supervised BSc and MSc work. Consequently, we embrace the university’s initiative to capitalise on its role as a centre for cutting-edge AI research with AI4all (AI in all TU Wien curricula). Accordingly, TU Wien recognises the omnipresence of algorithms and the increasingly central role of AI in our daily lives. These rapid transformations must be adequately reflected in TU Wien’s curriculum and research programs. This does not only secure the student’s competitiveness in the job market and the university’s research excellence but also enhances the benefits of AI for society, and safeguards us from the adverse consequences for privacy, ethics, and transparency. However, to adequately address all faculties’ demands, we require more extensive research on methods for AI in science and engineering. In addition, we need to assist other faculties in utilising the available internal computational resources, by reducing the barriers of programming basics and practical data science. In our experience, web-based interfaces have been instrumental in teaching simulation-based research and data-driven discovery in both classroom activities and hands-on programming exercises. We therefore anticipate that the TU Wien cookbooks would be beneficial for the implementation of a variety of teaching and research activities across all faculties. They provide a canvas on which domain specific knowledge can be combined with the functional expertise needed to fully exploit the benefits of AI: Allan Hanbury & Nikolas Popper (Institute of Information Systems Engineering)

Faculty of Civil and Environmental Engineering¶

The research field of civil and environmental engineering is in flux, where there is a growing move to digitize and more computer-oriented methods are developed. For example, digital twins, digital representations of buildings and cities, are the next evolution of civil engineering. Such developments are a driving force behind a completely new master study in civil engineering, which has been dubbed Digital Civil Engineering Science (or “DiCES”). While this MSc program continues to rely on the fundamentals of structural analysis—still grounded in mathematical principles and mathematics in chalk—it advances traditional methods by integrating computer science. This broadens students’ skillsets with numerical methods essential for optimization techniques and simulations that refine and advance building processes. The implementation of notebooks is then also foreseen for this new MSc program: Sabine Sint (Institute of Material Technology, Building Physics, and Building Ecology)

Student skill levels and teaching goals¶

Faculty of Mathematics and Geoinformation¶

In 2022, I launched a new course at the Department of Geodesy and Geoinformation titled Introduction to Programming for Geodesy, Geoinformation and Environmental Engineering. The goal was to equip students with foundational Python skills, progressing from basic concepts to practical applications in Earth sciences. The course was entirely delivered through Jupyter Notebooks within the dataLAB JupyterHub — a platform specifically designed to offer students low-barrier access to an e-learning environment. Despite our efforts to keep the material accessible through basic examples, we quickly realised we had overestimated the students’ technological readiness and computational thinking skills. More than 50% of the participants struggled to keep up. After discussions with colleagues from other faculties (Informatics, Civil Engineering, Chemistry, Mechanical Engineering), it became clear that this issue extended beyond our own department and faculty. In response, we joined forces to develop a more general and interdisciplinary entry level course: Python for Natural Sciences. Looking back from the vantage point of 2025, this initiative proved to be a crucial step. The course now attracts a wide range of students, including those from data science and informatics, physics, regional planning, architecture, etc., and has become a cornerstone for fostering digital literacy across the sciences. Our experience underscores a vital lesson: in an era of rapid digitisation and the growing influence of AI in the workplace, strengthening students’ digital competencies is not just beneficial — it’s essential : Gottfried Mandlburger (Department of Geodesy and Geoinformation)

Faculty of Electrical Engineering and Information Technology¶

Employing Jupyter notebooks for teaching requires an acute awareness of the digital literacy level of our students. Notebook can be a medium to teach students how to interact with data and distil actionable information to, for example, disclose flow patterns in networks and simulate network attacker scenarios. Nevertheless, notebooks might not be suitable in every learning environment and might even be harmful for teaching advanced programming skills. This, as notebooks can lead to bad coding practices, which are detrimental for production ready code projects. An example of the downside of Jupyter notebooks is the non-linear workflow, where an unwitting student can run code chunks out of order. Furthermore, it might cause students to write messy code; not modular/structured, excessively sequential, and abuse global variables. In the process the student might lose skills related to strategic, hierarchical, structured and organised thinking. It is therefore of importance to carefully consider the learning environment in which to employ notebooks. In addition, when opting for notebooks, then we should promote best practices for coding; e.g., meaningful variables names, consistent formatting, and modular code chunks: Tanja Zseby & Felix Iglesias (Institute for Telecommunication) Implementation of notebooks in teaching

Faculty of Technical Chemistry¶

Teaching about the state of the art of your scientific discipline is integral to academic education. It highlights the importance of your field of research and thereby encourages participation and performance in the classroom. In the case of chemistry, teachings involve the use of machine learning, deep learning and artificial intelligence to study protein designs. However, today’s computationally driven research can make the translation from innovation to up-to-date teaching materials more difficult. In my experience, teaching about AI applications in chemistry is hard to imagine without Jupyter Notebooks and the support of TU Wien’s data lab. Data-driven storytelling and notebook-based teaching improves the student’s insight of the inner working of machine learning. In particular, ad hoc formulations of the basic building block of machine learning techniques in front of the classroom are a good way to captivate students by breaking down complex workflows in easy-to-understand chunks and by showing the thought processes of a researcher in action. These interactive lessons therefore serve as a demonstration of the flexibility and creativity involved with computational thinking—a valuable side-effect of notebook-based teaching: Esther Heid (Institute for Materials Chemistry)

Faculty of Physics¶

Teaching with notebooks has helped me better activate and motivate students during lectures. Especially, combining mathematical notations and running a code block that implements the previously introduced equation improves the student’s engagement. Such an approach can for example be used to visualize atomic orbitals through simulations with the Schrödinger equation. Students are then free to play around with the notebooks, where they can change the input for the variables and assess its impact through visual representations. This supports students who have a wide range of experience and abilities. Students needing more support find the scaffolding and incremental introduction of equations and numerical solutions a comforting way to ease into a topic. Whereas more advanced students can experiment with the provided code to find their own solutions and develop new questions. As often the case, there are different ways to analyse and approach things—the notebook is just a starting point. By playing with the notebooks, new insights gained by the student might be a catalyst for independent research projects, such as thesis work: Markus Valtiner (Institute of Applied Physics)

Data and compute Resources¶

Faculty of Mechanical and Industrial Engineering¶

Programming skills are essential for mechanical engineers, as they are required for running simulation and performing data analysis. Introductory courses on the basics of programming, technical drawing, and numerical methods address these requirements and help the student analyse turbulence data and perform optimization tasks. Problematic is, however, using notebooks to reach these teaching goals, as typical simulation experiments of turbulent flows can generate data that range in the Terra-bytes: Alfredo Soldati (Institute of Fluid Mechanics and Heat Transfer)