This study examines the experiences, perceptions and difficulties encountered by secondary school students while interacting with SGAMES21 mini, an Intelligent Tutoring System (ITS) designed to support critical thinking development through environmental problem-solving activities. The research addresses a persistent gap in the literature concerning students’ subjective engagement with AI-driven learning tools, particularly within under-resourced educational contexts. Fifty eighth-grade students from five public schools in Montería, Colombia, participated in the study. A qualitative, exploratory approach was employed, with data collected through focus group interviews and the analysis of visual artifacts created by students following their use of the ITS.

The instructional task was organized using a three-phase cognitive framework—information analysis, evaluative reasoning, and judgment formulation—embedded in a narrative focused on deforestation. The findings demonstrate the relevance and novelty of combining narrative-driven AI tutoring with critical thinking instruction. Regarding their learning experiences, students reported high levels of engagement and expressed that the system enabled them to explore real-world dilemmas through guided reflection. In terms of perceptions, participants articulated a sense of autonomy and confidence in decision-making, often referencing improvements in argument construction and ethical reasoning. Despite these positive outcomes, students also faced difficulties related to task interpretation, information structuring, and technical interface responsiveness.

The contribution of this research lies in advancing the understanding of how ITS environments can be designed to support higher-order cognitive processes in diverse learning settings. It offers empirical evidence that advances pedagogical and technological models for AI-mediated education. The implications suggest that while gamified and interactive features enhance motivation, future systems should incorporate adaptive scaffolding and clearer instructional guidance to optimize learning outcomes.

In addressing a defined research gap, this study not only enriches current discourse on human–AI interaction in education but also outlines future directions for system design and policy implementation aimed at fostering critical thinking in marginalized school populations.

Keywords: Intelligent Tutoring Systems (ITS), Critical Thinking, Secondary Education, Artificial Intelligence in Education, Human-Computer Interaction, Metacognition, Educational Technology Design.