Enhancing student confidence in data processing through ChatGPT-assisted coding in chemistry education
Bibliographic Data
| ID | 22436592 |
|---|---|
| Authors | Calum K Gordon (0000-0002-4963-1896, Massey University, corresponding author), Lara D Browne (Victoria University of Wellington), Tylah G Sweet (Victoria University of Wellington), Chase Zemke-Smith (Victoria University of Wellington), Finnian J Blaauw-Smith (0009-0008-0353-5314, Victoria University of Wellington), Sanutep V Chan (0009-0000-0282-9670, Victoria University of Wellington), Mark Waterland (Massey University), Nathaniel J L K Davis (0000-0003-2535-8968, Victoria University of Wellington, corresponding author) |
| Year | 2026 |
| Volume | 5 |
| Issue | 1 |
| Publication date | 2026-04-16 |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | Discover Education (JOURNAL) |
| Journal identifiers | ISSN: 2731-5525 • E-ISSN: 2731-5525 |
| Publisher | Springer Science and Business Media LLC (PUBLISHER) |
| DOI | 10.1007/s44217-026-01479-w |
| OpenAlex | W7154622736 |
| Language | EN |
| References cited | 38 |
Large language models such as ChatGPT are steadily becoming more prevalent in both chemistry and chemical education. These models are not only easy to access, but are invaluable tools for writing, research, question-answering, and code generation. The seamless integration of large language models into the field of chemistry is seemingly inevitable – and so it is important to outline the benefits of their use as well as their limitations. Herein, we present a positive use of ChatGPT to improve participants confidence in data processing skills. Participants were asked to sign up for the workshop, and a total of 32 participants underwent the activity and subsequent survey. The exercise aims to teach participants how to interface with ChatGPT to generate tailor-made code for data analysis of large raw datasets. Participants were tasked with prompting ChatGPT with code that allows them to manipulate a provided raw dataset into figures and plots for easy interpretation. The exercise resulted in short-term confidence gains, with 87% of participants indicating they intend to use ChatGPT to assist with data processing in the future, and ~ 97% of participants indicating neutral to very good confidence in data processing with code post-activity (32 participants). While no control group or long-term follow-up was included, the results indicate short-term gains via the use of ChatGPT, alongside identifiable limitations related to prompting and model errors.
Confidence interval · Data Processing · Raw data · Advanced Electron Microscopy Techniques and Applications · Artificial Intelligence in Healthcare and Education · Machine Learning in Materials Science
| Highly cited | No |
|---|