Artificial human thinking: ChatGPT’s capacity to be a model for critical thinking when prompted with problem-based writing activities

Abstract

Abstract The ability to think critically is an important and valuable skill that students should develop to successfully solve problems. The process of writing requires critical thinking (CT), and the subsequent piece of text can be viewed as a product of CT. One of the strategies educators may use to develop CT is modelling. Given ChatGPT’s ability to generate textual responses, one could suggest that educators can use AI-generated responses to problem-based writing activities as a model for CT. This study employed a qualitative research approach guided by the interpretivist paradigm to thematically analyse 57 digital texts produced by ChatGPT 3.5. The study found that ChatGPT 3.5 has the capacity to produce a model response which exhibits CT skills and dispositions to three different degrees namely explicit, guided, and limited. Each degree requires a certain level of educator involvement for ChatGPT 3.5ʹs textual responses to be used as a model for CT.

Keywords

• Critical thinking
• Prompt engineering
• ChatGPT
• Large language models
• Artificial intelligence
• Critical thinking skills