Navigating cognitive dissonance: master's students' experiences with ChatGPT in dissertation writing.

With the increasing prevalence of AI tools like ChatGPT in academic settings, understanding their impact on students' psychological experiences during dissertation writing is crucial. This study aims to explore the cognitive dissonance experienced by master's students during dissertation writing with the assistance of ChatGPT and identify the strategies they employ to manage this dissonance. Using grounded theory as the primary research methodology, we analyzed 28 interview transcripts to uncover key elements of cognitive dissonance and develop a corresponding theoretical model. Our findings revealed that the primary sources of cognitive dissonance among master's students were the strong intentions to use ChatGPT driven by subjective norms and technological expectations, conflicting with the reality of multiple choices. To alleviate this cognitive dissonance, students adopted strategies such as improving prompt quality, feeding relevant domain-specific data to the AI, avoiding academic misconduct, and maintaining academic integrity. This study challenged and extended the Technology Acceptance Model and the Theory of Planned Behavior by incorporating cognitive dissonance, and emphasized the underlying pathways and causes of dissonance. Practically, our findings offer significant implications for institutions and educators, emphasizing the importance of supporting and guiding the use of generative AI tools like ChatGPT in dissertation writing.