OBJECTIVE: To assess whether or not university instructors can differentiate reflections generated by ChatGPT from those generated by students, and to assess whether or not the content of a thematic analysis generated by ChatGPT differs from that generated by qualitative researchers on the same reflections.
METHODS: Hardcopies of 20 reflections (10 generated by undergraduate dental students and 10 generated by ChatGPT) were distributed to three instructors who had at least 5 years of teaching experience. Instructors were asked to assign either \'ChatGPT\' or \'student\' to each reflection. Ten of these reflections (five generated by undergraduate dental students and five generated by ChatGPT) were randomly selected and distributed to two qualitative researchers who were asked to perform a brief thematic analysis with codes and themes. The same ten reflections were also thematically analyzed by ChatGPT.
RESULTS: The three instructors correctly determined whether the reflections were student or ChatGPT generated 85% of the time. Most disagreements (40%) happened with the reflections generated by ChatGPT, as the instructors thought to be generated by students. The thematic analyses did not differ substantially when comparing the codes and themes produced by the two researchers with those generated by ChatGPT.
CONCLUSIONS: Instructors could differentiate between reflections generated by ChatGPT or by students most of the time. The overall content of a thematic analysis generated by the artificial intelligence program ChatGPT did not differ from that generated by qualitative researchers. Overall, the promising applications of ChatGPT will likely generate a paradigm shift in (dental) health education, research, and practice.
目的:评估大学教师是否可以将ChatGPT产生的反射与学生产生的反射区分开来,并评估ChatGPT生成的主题分析的内容是否与定性研究人员在相同的反射上生成的内容不同。
方法:将20个反思的硬拷贝(10个由本科牙科学生生成,10个由ChatGPT生成)分发给三名具有至少5年教学经验的教师。要求教师为每个反射分配\'ChatGPT\'或\'student\'。随机选择其中的十个(五个由牙科本科生产生,五个由ChatGPT产生),并分发给两名定性研究人员,他们被要求对代码和主题进行简短的主题分析。ChatGPT也对相同的十个反射进行了主题分析。
结果:三位教师正确地确定了反射是学生还是ChatGPT产生了85%的时间。大多数分歧(40%)发生在ChatGPT产生的反射中,作为教师认为是由学生产生的。将两位研究人员产生的代码和主题与ChatGPT产生的代码和主题进行比较时,主题分析没有实质性差异。
结论:教师可以区分ChatGPT或大多数时候学生产生的反射。人工智能程序ChatGPT生成的主题分析的总体内容与定性研究人员生成的主题分析的总体内容没有差异。总的来说,ChatGPT的有希望的应用可能会在(牙科)健康教育中产生范式转变,研究,和实践。