<p>Maintaining student motivation is a persistent challenge in modern education, as traditional methods often fail to fully address students’ individual interests. While personalized approaches show promise, educators face substantial challenges in creating personalized educational content due to the significant resources required, often exceeding practical limitations. Context-personalization through Generative Artificial Intelligence (GAI) enables the real-time creation of diverse learning material and tasks tailored to each student’s unique interests, surpassing what educators can provide in practice. This paper reports on the development of such a GAI-based tool and its systematic investigation. A 2<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\times \)</EquationSource> <EquationSource Format="MATHML"><math> <mo>×</mo> </math></EquationSource> </InlineEquation>2 mixed between-subjects design (<i>N</i>&#xa0;=&#xa0;114) was conducted in primary school classrooms in Germany during regular math classes, with learning material (context-personalized vs. standard) and task type (context-personalized vs. standard) as factors. Results show that learning with context-personalized learning material and tasks led to higher intrinsic motivation, greater interest, and higher learning performance compared to standard learning material and tasks. The findings indicate that interests outside of typical school subjects can trigger situational interest and enhance intrinsic motivation in academic contexts. Furthermore, this paper highlights cutting-edge technologies, showcasing how Large Language Models, such as GPT-4, can place fractal text tasks in meaningful contexts in real time. This capability provides extensive opportunities to create learning material and tasks that foster intrinsic motivation, interest, and improve learning performance, effectively addressing the limitations previously faced by teachers.</p>