Document is current

Article History

Received: 15 June 2025

Accepted: 28 August 2025

First Online: 1 September 2025

Declarations

:

: Not applicable.

: Not applicable.

: Use of large language models (LLMs) in statistical calculations and visualization is described in Sect. “”. During multiple stages of this study ChatGPT-4o, ChatGPT-4.5, ChatGPT-o1 (April 2025), was employed as a coding assistant to develop Python scripts for data processing and visualization. Code was generated iteratively in collaboration with the authors via Google Colab, and targeted tasks included cleaning and structuring Excel-derived datasets, computing descriptive and inferential statistics (e.g., Spearman’s ρ), and rendering figures (e.g., Bloom category distributions, institutional typologies). All AI-generated code was manually reviewed, edited, and executed by the authors to ensure computational validity, reproducibility, and alignment with the study’s analytical aims. No statistical inference or interpretive output was directly accepted without verification. Additionally, ChatGPT was used for linguistic polishing, reformulation and reorganization of paragraphs for clarity, and condensation of dense sections, without altering the original argumentation, epistemic stance, or analytical conclusions. The final manuscript reflects the authors’ conceptual design, analytic decisions, and scholarly interpretation. All AI contributions were non-autonomous and subject to critical human oversight.

: The authors declare no competing interests.