Generative artificial intelligence has entered higher education with force, but its incorporation into university learning still raises a key question: how can a tool such as ChatGPT be turned into a genuinely useful teaching resource, rather than simply a shortcut for obtaining answers?
The article “Generative artificial intelligence for self-learning in higher education: Design and validation of an example machine” addresses this issue through a specific proposal: the design and validation of an example machine intended to help university students review Educational Research Methodology content through practical cases, guided interaction, and immediate feedback.
One of the most interesting aspects of the study is that it shows that working with AI requires pedagogical design, not just technological enthusiasm. The authors develop different prompts, validate them with expert judges, and test them with 192 students in the field of Education Sciences.
The results point to high levels of satisfaction and perceived usefulness, but they also reveal something especially relevant: simpler prompts work better than more complex ones. Far from being a minor detail, this finding reminds us that generative AI needs boundaries, clear instructions, and well-defined tasks in order to become an ally of autonomous learning.
The study is also valuable because it does not hide the tool’s errors; on the contrary, it turns them into part of the analysis. In fact, students who witnessed failures during the pilot study rated the usefulness of the machine more positively, perhaps because those errors made visible the need to look critically at generated responses.
This is one of the article’s key insights: AI can support self-learning, but its greatest educational potential does not lie in providing perfect answers, but in encouraging practices of analysis, verification, and decision-making. Rather than presenting AI as a closed solution, the study opens up a promising line of work for designing university learning experiences in which learning with technology also means learning to distrust it intelligently.
---
How to Cite: Sánchez-Prieto, J. C., Izquierdo-Álvarez, V., del Moral-Marcos, M. T., & Martínez-Abad, . F. (2025). Generative artificial intelligence for self-learning in higher education: Design and validation of an example machine. RIED-Revista Iberoamericana de Educación a Distancia, 28(1), 59–81. https://doi.org/10.5944/ried.28.1.41548
