A Comparative Study of the Efficacy of LLMs in Generating Learning Resources for Non-English Major Postgraduates

This study examines the efficacy of prominent large language models, ERNIE Bot and ChatGPT, both domestically and internationally, in producing pertinent language learning materials. The research studied variations in the outcomes of two large language models in assisting non-English major postgraduate students at a university in generating supplementary learning resources. The materials produced by ERNIE Bot and ChatGPT were assessed across four dimensions: vocabulary, translation, reading, writing with evaluations solicited from students and educators. By analyzing the discrepancies in the outputs of the two large language models in producing educational resources, valuable insights were garnered regarding their inherent strengths and weaknesses. An examination of the underlying factors contributing to these differences was conducted, with the ultimate goal of augmenting the generative competencies of large language models, offering more varied developmental pathways for the application of artificial intelligence in education.