Assessing the Potential of Large Language Models for Chemical Engineering Applications

This study investigates the emerging potential of large language models in chemical engineering by evaluating GPT-4's performance on questions from general chemistry, physical chemistry, and reactor and process design. GPT-4 correctly answered 45% of the problems without error, with overall accuracy increasing to 62% when partially correct answers were included. These results highlight both the capabilities and limitations of LLMs in this field. We discuss the challenges and future prospects of integrating LLMs into chemical engineering, aiming to provide insights and direction for future interdisciplinary research. Additionally, a problem set is provided for further exploration.