Comparing the Characteristics of English Compositions by Korean EFL College Students and AI Tools

The purpose of this study was to find out the characteristics of outputs produced by artificial intelligence-based tools and English compositions written by English learners, in terms of general metrics, and accuracy and clarity, respectively. A total of 10 compositions (five students’ English compositions and five artificial intelligence produced outputs) submitted by five students were quantitatively analyzed by using artificial intelligence-based writing evaluation tools such as Wordcounter and Grammarly. In addition, the collected compositions were compared and qualitatively analyzed sentence by sentence by the researchers. The results of the study analysis are as follows: 1) There was no significant difference in the basic characteristics of the students’ English compositions and artificial intelligence produced outputs, but the qualitative analysis results at the sentence level showed differences in terms of vocabulary diversity and depth; 2) In terms of accuracy, the students’ English compositions had significantly more errors than the artificial intelligence produced outputs, and in terms of clarity, the artificial intelligence produced outputs had more errors. Base on the research findings, educational implications for the practical implementation of artificial intelligence tools in the EFL classroom were suggested.