Project-based learning to teach GenAI-assisted biogeochmical model development.

Through teaching the fundamentals of coding environmental models, we contribute to the development of computational competencies within the study programmes of D-USYS. Through this proposal, we aim to improve the course in three main ways.

Firstly, with the emergence of generative AI (GenAI), traditional coding exercises alone are no longer sufficient to prepare students for workplaces in which numerical tasks are efficiently supported by GenAI. Therefore, we will develop a semester-long group project that integrates GenAI as a coding aid while emphasizing project-based and self-directed learning. At the same time, we will safeguard the acquisition of fundamental coding skills by prohibiting the use of GenAI during the first four lectures.

Secondly, while the course currently primarily targets students from the Master in Agricultural Sciences, we will adapt the programming examples to make them more attractive and relevant to students in the Master in Environmental Sciences.

Thirdly, we will evaluate the suitability of the “ETH Code Expert” platform and the “Positron” integrated development environment (IDE) as new modelling tools within the course. “ETH Code Expert” will allow us to centralize coding-related exercises, solutions, and feedback in a single online environment, supported by the developers of the platform at ETH. “Positron”, which is optimized for both R and Python, offers seamless integration of GenAI and agentic coding into the programming workflow.

The proposed changes will be implemented in FS2027 and will lay the foundation for a future expansion of the course from 3 to 6 ECTS in a subsequent development phase.

• Teaching the development and implementation of biogeochemical models to answer research questions in a
  semester-long project, integrating the use of GenAI.
• Using student projects as a tool to integrate theoretical knowledge and practical application in a real-world
  mimicking setting under low stakes.
• Preparing students for a GenAI-powered project-based working environment with a semester project that blends
  project-based learning with GenAI use and allows us to give personalized guidance.
• Strategically positioning the course as one of the main MSc. courses at D-USYS that teach computational
  competencies with GenAI integration.

Students:

• The course will become attractive to students from a broader range of study programmes.
• Project-based learning will foster self-directed and active learning.
• Exposure to GenAI will strengthen AI fluency and computational competencies.
• Newly introduced skill-transfer exercises will support the application of acquired knowledge and skills.
• The “ETH Code Expert” platform will provide a centralized environment for exercises, progress tracking, and feedback.
• The “Positron” IDE will enable the seamless integration of GenAI into the programming workflow.

Lecturers:

• We will expand our pedagogical toolkit by adapting teaching strategies to GenAI-supported workflows (e.g., through “Positron”), implementing process-oriented assessment formats, and integrating the “ETH Code Expert” platform into the course.
• The proposed innovations will lay the foundation for a future expansion of the course from 3 to 6 ECTS within the framework of the PAKETH reforms.

Programme:

• The redesign will serve as a structured example for the responsible integration of GenAI into STEM education, generating insights that support the revision of the Environmental and Agricultural Sciences BSc and MSc curricula as part of the “Studiengangsinitiative² (SI²)”.
• Integrating Environmental Sciences into the course will create a more interdisciplinary learning experience and broaden the course’s appeal across programmes.
• The computational competencies, GenAI-related skills, and project implementation experience developed in this course are highly relevant for both today’s professional environment and scientific research.