Bridging the programming gap: Integrating AI-mediated coding tools to strengthen hydrometeorological data analysis competencies in undergraduate water science students

The rapid expansion of Generative Artificial Intelligence (GAI) offers unprecedented opportunities to support higher education, particularly in data-intensive scientific disciplines. In undergraduate water sciences programmes—such as the BSc in Water Resources from Rey Juan Carlos University (Spain)—students are required to interpret hydrometeorological datasets, understand the dynamics of the hydrological cycle, and apply analytical methods to real environmental problems. However, many students face persistent barriers when working with programming languages such as R (R Core Team, 2023), which are essential for exploring, processing, and visualizing hydrometeorological data. These limitations hinder their ability to achieve key learning outcomes related to data literacy, problem‑solving, and digital competence. To address these challenges, an educational innovation project was implemented in the core Hydrometeorology course (2^nd year) using a structured pedagogical strategy that integrates GAI as a learning support tool. The initiative combines: (1) teacher training through institutional AI‑literacy programs; (2) the redesign of practical activities to incorporate AI-mediated code generation; (3) explicit instruction on ethical, critical, and responsible AI use and correct prompt writing; and (4) the deployment of student surveys and performance analytics to evaluate effectiveness.

All the students completed the AI ethics training module and were also taught how to craft effective prompts, including strategies for specifying context, defining constraints, and iterating queries, to obtain accurate, reproducible, and pedagogically relevant outputs from GAI tools. AI tools—primarily Microsoft Copilot (Microsoft, 2025; https://copilot.microsoft.com/)—were used to scaffold R programming tasks linked to open hydrometeorological datasets from the Spanish Meteorological Agency (AEMET; opendata.aemet.es). This enabled students to focus on conceptual understanding rather than syntactic details. Student perceptions were assessed through structured surveys, and academic performance was contextualized by comparison with by comparison with repeat students who had previously taken the course without AI support. Preliminary results from two structured student surveys (N=12) indicate positive perceived impacts. Among first‑time students, 7/9 reported increased autonomy in hydrometeorological data analysis, 8/9 found the AI support pedagogically useful, and 8/9 recommended its continued use. Recurrent students who had previously taken the course without AI reported reduced perceived difficulty and improved performance, with all of them acknowledging higher confidence when working with data and code.

This project provides empirical evidence on how GAI acts as a pedagogical scaffold to reduce programming barriers, foster inclusive learning, and enhance motivation. By improving data analysis competencies, it supports Sustainable Development Goals on education, water management, and climate action, while informing future curricular innovations in Earth and Environmental Sciences programmes. The authors acknowledge the support of the Vice-Rectorate for Academic Innovation of Rey Juan Carlos University through the 2025 Teaching Innovation Project Call, which made this initiative possible.