2025 AIChE Annual Meeting

(262g) How Shall Large Language Model Apply to University Lab Teaching?

Authors

Wei Sun - Presenter, Beijing University of Chemical Technology
The surge of Large Language Model (LLM) cannot be ignored everywhere in our life. All activities in universities are under its impact, no matter the positive or negative feedback is given. Students are group of people who can easily get new ideas and adapt new tools in their life. The difference is that LLM has a big potential to change the way they learn in university and perhaps their career life in future. As a professional educator in university, to get a positive involvement will be better to try with students together, therefore to identify the impact and try to come up with a positive strategy to survive and eventually lead the revolutionary change.

A junior lab course is taken as an example in this trial. Students participate voluntarily. A possible list of LLM benefit to the course was created by brain storm. A systematic test has been conducted according to the list. Experiments in the lab include six applications from chemical engineering course before junior year. In order to prepare, conduct the experiment and finalize the report, students need to get all related knowledge together, which should be available in their previous text books and course materials, but scattered in different courses, which include but not limited to fundamental theories and equations. At the meantime, they also need to follow certain protocol for scientific writing. During the experiment, chemical data and equipment operation procedure are also needed. Traditionally they can be obtained by library searching but more time consuming. With LLM available, there is an alternative way to do it more efficiently. The only concern is that the reliability of information provided by LLM. The results will always fall into two categories, positive and negative. Negative impact causes more concern from both students and instructors. The question is if it is possible to further train the LLM with domain knowledge to improve it. Given the interactive nature of communication with LLM, is there a better way to get the help from LLM.

To answer all questions above, a group of students conduct the experiment together after the semester with the aim to provide the initial trial to our community. Hopefully it can signal the pitfalls on the way of new era for LLM in university learning. Final report is on the way.