This study set out to determine whether an artificial-intelligence-driven personalized curriculum can raise mathematics and English achievement among visually-impaired junior-secondary learners in low-resource public schools in Imo State, Nigeria. The prevailing problem is that static Braille textbooks arrive late and impose uniform pacing, thereby widening achievement gaps. Seventy-two learners with best-corrected visual acuity of 6/60 or worse and no additional intellectual disability were recruited from three special schools representing urban, peri-urban and rural zones. Using a mixed-methods quasi-experimental design, participants were randomly assigned to the VI-AI group (n = 36) or the Braille-control group (n = 36). Over twelve weeks the VI-AI group studied on low-cost Android tablets that adapted difficulty, provided speech/haptic feedback and exported content to braille displays, while the control group followed identical objectives through conventional Braille textbooks. Pre-, post- and delayed tests were supported by usability and self-efficacy scales plus semi-structured interviews. ANCOVA revealed a large post-test advantage for the VI-AI group (d = 0.81, p < .001) and a medium retention effect four weeks later (d = 0.48). Usability exceeded the acceptability threshold (78.3/100) and digital self-efficacy rose by 0.9 scale points (d = 1.05). Qualitative themes linked the gains to learner autonomy but also highlighted power outages as a threat to retention. The findings indicate that AI-driven personalization can significantly enhance academic performance and perceived inclusion of visually-impaired learners, provided that scale-up is accompanied by stable power, targeted teacher professional development and open-source adaptive engines.
Eke et al. (Thu,) studied this question.