In recent years, large language models have made significant progress in the field of language processing, but they face challenges in molecular optimization for drug discovery. However, researchers have successfully achieved real-time human-computer interaction in the molecular optimization process through the development and application of the DrugAssist model. This model excels in single-attribute optimization and demonstrates excellent transferability in zero-shot and few-shot scenarios, offering the possibility of real-time interaction and iterative optimization for drug discovery.