FermentAI: Large Language Models in Chemical Engineering Education for Learning Fermentation Processes

Recent developments in artificial intelligence (AI), leading to the release of continuously improving large language models (LLMs) provide the opportunity for educators to automate repetitive tasks with stimulant experiences for the students. In fact, the ability of LLMs to extract content and key information from text offers a powerful tool for enhancing the learning experience. In this work, we present an example of how LLMs can be used to automate educational processes. We implement FermentAI, a virtual tutor (VT) to answer students' questions about fermentation for a Master's Degree course taught at the Technical University of Denmark. The model used is a pre-trained sequence-to-sequence model. The prompt provided to the LLM is composed of a question and its context, the curated information for the model to generate the answer. The context is retrieved through semantic search by calculating the cosine similarity between the query question and a dataset of historical questions. The primary objective of this work is to create an interactive, freely available, and accurate tool that students can use to ask questions about fermentation. The VT is integrated into BioVL, an online educational platform for (bio)chemical processes developed by the authors. The code and data are open-source on GitHub.