Emily F. Ruff, Mark A. Engen, Jeanne L. Franz, Jonathon F. Mauser, Joseph K. West, Jennifer M. O. Zemke
Large language models (LLMs) such as ChatGPT have recently been challenging traditional models of higher education. Given the growing use of these tools for writing, research, and content retrieval tasks, it is imperative that both students and faculty understand their capabilities and shortcomings. Here we describe assignments in which lower- and upper-division students evaluated chemistry writing samples generated and revised by ChatGPT version 3.5 and used this program for revision and other writing tasks. General Chemistry students who evaluated AI-generated content showed strong gains in their knowledge about report structure and the capabilities and deficiencies of LLMs in chemistry. Upper-division students found generative AI to be helpful for revision. Content analysis revealed AI-revised and -generated samples exhibited fewer grammatical errors, fewer very short and very long sentences, and improved readability of the text. We conclude with some implications for future research and suggestions for other instructors who wish to use and adapt these assignments.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados