Experiment: Leiden University student writes thesis with just AI tools for supervision. “It requires a high achieving, very independent student who is able to think critically.”

Exploring the Future of Academic Supervision: A Leiden University Student’s Thesis Crafted with AI Assistance

In an innovative experiment at Leiden University, a student recently completed their thesis primarily utilizing artificial intelligence (AI) tools for supervision and guidance. This approach offers a glimpse into the evolving landscape of academic research and highlights both the potential and limitations of AI in higher education.

The student undertaking this project emphasized that such a method requires a highly capable and independent learner—someone who can think critically and navigate complex topics without extensive traditional oversight. This initiative underscores the importance of self-motivation and analytical skills, which are essential to effectively leverage AI tools without compromising academic integrity.

Interestingly, while there may be students engaging in similar practices discreetly, the openness of Leiden University’s supervisors to this experimental approach marks a significant shift in academic pedagogy. By embracing AI-assisted research, educators can explore its effectiveness across different facets of thesis development. Preliminary observations suggest that AI can be particularly useful in initial research phases, literature reviews, and drafting assistance. However, challenges remain in areas such as critical analysis, nuanced interpretation, and ensuring the originality and accuracy of references—especially considering AI’s current propensity to generate plausible but sometimes fabricated sources.

The experiment also highlights the importance of student selectivity; only those with strong critical thinking skills and academic discipline are well-suited for such an approach. This ensures that the resulting work maintains scholarly rigor and does not fall into the trap of superficial or sloppy writing driven solely by AI-generated content.

Regarding grading, the completed thesis received an approximate score of 8.5 on the Dutch evaluation scale, which correlates to a B+ or potentially an A grade (above 3.7 GPA based on the Oppenheim or similar scales). This indicates that, with appropriate oversight and critical engagement, AI-assisted research can lead to high-quality academic outputs.

This experiment at Leiden University points toward a future where AI becomes a complementary tool in academic research and supervision, provided it is used judiciously by capable students. As higher education institutions continue to explore these boundaries, ongoing research and dialogue will be essential to establish best practices and standards for AI’s role in scholarly work.