How do pre-service language teachers perceive generative AIs’ affordance?:A case of ChatGPT
Yuko Goto Butler, Shiyu Jiang
Abstract
As generative artificial intelligence (AI), such as ChatGPT, becomes prevalent, it presents various ethical and practical challenges for schools. This study explores how students assess ChatGPT's affordances and limitations in academic writing and how these assessments relate to their writing performance, judged by themselves and the course instructor. The study uniquely focuses on international pre-service teachers in a Teaching English to Speakers of Other Languages program in the United States ( N = 38), who may use generative AIs in their teaching and studies. The participants completed tasks in an applied linguistics class, including writing essays with and without ChatGPT on the same topic and comparing the two essays. Grounded in the Theory of Affordance, the students' evaluations were coded and analyzed with Epistemic Network Analysis. Results showed that students viewed ChatGPT's affordances and limitations differently according to their writing profiles. Those with lower writing scores or confidence were less critical of ChatGPT, highlighting its merits in language-related elements such as grammatical accuracy and lexical choice. The study suggests the importance of assisting future language teachers in fostering accurate knowledge of AI's affordances and developing balanced views towards AI through professional development, while considering individual differences in their needs and abilities.