Background <p>UNESCO reports that around 70 countries have adopted AI-related strategies, recognizing AI literacy as essential for preparing citizens in an AI-driven world. Yet, two key challenges remain: limited AI literacy development at the foundational level and persistent gender gaps in AI fields. Without early, inclusive education, students—especially girls facing STEM-related barriers—may lack the skills and confidence to engage with AI. These challenges are interconnected: focusing on AI literacy without gender equity may reinforce gaps, while promoting inclusion without strong AI foundations may limit impact. This study revised an AI literacy questionnaire for fifth-grade students, covering affective, behavioral, cognitive, and ethical dimensions, and examined whether gender disparities exist and whether AI and AI literacy courses can help reduce them.</p> Results <p>This study includes two parts. Study 1 revised and validated an AI literacy questionnaire for fifth-grade students (<i>N</i> = 504), showing strong reliability, validity, and gender invariance. Results indicated that male students scored higher than female students. Study 2 adopted a quasi-experimental design to compare two courses. The comparison group (<i>n</i> = 55) took an AI course on real-world Artificial Intelligence of Things (AIoT) applications, while the experimental group (<i>n</i> = 54) followed an AI literacy course integrating STEAM, project-based learning (PBL), and AIoT. Both groups completed pre- and post-tests. Results showed improvements in both groups. While female students scored lower on the pre-test, no gender differences remained on the post-test.</p> <p>However, the AI literacy course produced greater overall gains, while ethical outcomes remained similar across groups. Notably, female students in the AI literacy course outperformed male students in the AI course, indicating its potential to address gender disparities.</p> Conclusions <p>This study found that gender disparities in AI literacy emerge at the foundational level but can be reduced through targeted AI literacy instruction. A validated questionnaire assessing affective, behavioral, cognitive, and ethical dimensions was developed to support instructional planning. Findings support integrating AI literacy courses into elementary education through a flexible STEAM-, PBL-, and AIoT-based course model. This adaptable approach has the potential to scale across contexts, enhance engagement, support inclusion, and reduce the global AI talent gap.</p>