ChatGPT in PhD Mentoring: Exploring the Potential of Generative AI for Academic Guidance and Sustainable Educational Practices

While the applications of Generative AI in education and academic research are growing, its potential for supporting PhD-level mentoring and academic counseling remains underexplored. This exploratory study evaluates the relevance and appropriateness of ChatGPT-generated recommendations for PhD research, with a particular emphasis on its potential to promote sustainable and resource-efficient practices in doctoral education. Using a real-world case study on disaster risk management, input prompts were designed with varying levels of contextual detail, including naïve prompts, keywords selected by supervisors, keywords generated by ChatGPT, and topic-specific concepts derived from literature on advanced academic frameworks. The outputs were evaluated by a panel of five external academic experts who assessed their relevance, depth, and applicability to the research objectives. Further analysis explored tailored prompts designed to align with distinct research pathways. The results demonstrated that outputs based on topic-specific concepts received the highest appropriateness ratings, with strong interrater agreement. Naïve prompts also produced relevant outputs, while keyword-based prompts were rated lower, often failing to integrate core elements into cohesive recommendations. Tailored prompts reflecting specific research pathways were consistently rated as highly actionable and contextually grounded, highlighting ChatGPT’s ability to align with academic goals and contribute to sustainable educational innovation. These findings underscore ChatGPT's potential to complement PhD supervision by offering structured guidance, actionable insights, and timely feedback, paving the way for the "tripartite mentoring model," where AI collaborates with supervisors and students to address complex academic challenges in a sustainable and impactful manner.