Revisiting a Common Cobalt Chloride Equilibrium Experiment with Generative Artificial Intelligence

Abstract

We have developed and implemented a first-year undergraduate laboratory experiment at St. Bonaventure University that augments a well-established thermodynamics lab on the spectrophotometric determination of the equilibrium of cobalt(II) chloride hexahydrate with generative artificial intelligence (GenAI). Students utilized the aggregate of their own knowledge, peer discussion, instructor guidance, and free ChatGPT accounts in their analysis of laboratory data to iteratively refine their answers to chemical questions. Overall, the inclusion of GenAI in student workflow did not appear to negatively impact student performance, but there were instances where it struggled to produce correct answers for tasks. Descriptive analysis of student responses, chat logs, and instructor discussions suggests that most students improved their final answers despite potential shortcomings in GenAI responses. However, the analysis additionally identified multiple areas which could be improved in future iterations of this work. This laboratory experiment is designed such that it can be easily adopted by faculty with an interest in exploring the use of GenAI in chemistry curricula. In future work we intend to adapt the iterative framework and Sankey Flow diagrams employed in this work to design assessments which can disentangle the influence of peers, instructors, and GenAI on students’ work.