As society moves towards the intelligence era, the large language model ChatGPT has garnered widespread attention, particularly regarding its applications in education and the ethical issues it raises. This study examines the ethical dimensions of classroom dialogue for novice and expert teachers using a method that combines human verification with ChatGPT’s judgment. The content of classroom dialogue from expert teachers was used as prompts for ChatGPT to predict and generate subsequent classroom dialogue, which was then compared with those of the expert teachers. Results show that expert teachers exhibit significantly higher ethical levels in classroom dialogue compared to novice teachers. While there are differences in the characteristics of dialogue content predicted by ChatGPT and those of expert teachers, there is no significant difference in ethical levels between the two types of dialogue.