Improving Learner Performance Through Score Based Analysis on a Gamified AI Driven Cybersecurity Learning Platform

Abstract This study examines whether structured engagement with instructional content improves learner performance on Cyber Security Academy, an AI-driven, gamified cybersecurity learning platform. Activity logs covering 1,500 completed assessments from users in ten Anglophone and Francophone countries were analyzed by comparing learners who consulted the lessons and completed the full learning pathway (n = 900) with those who engaged only partially with the platform (n = 600). Group differences were tested with Welch's t-test, the Mann Whitney U test, one-way ANOVA, Spearman correlations, and Cohen's d as a measure of effect size. Learners who read the lessons before assessment obtained a mean score of 71.27 % versus 60.41 % for non-readers (Δ = +10.86 points; t(1192) = 12.21; p < 0.001; d = 0.650; 95 % CI [9.12, 12.60]), committed fewer errors (2.50 vs. 4.85; d = -1.069), and responded faster (9.85 vs. 12.69 minutes; d = -0.807), indicating greater fluency of cognitive processing. The effect decreased monotonically with age (d = 1.205 for learners aged 13 and under; d = 0.387 for those 22 and over) and with digital literacy (d = 1.434 for beginners; d = 0.524 for experts), while remaining positive in all ten countries. These findings indicate that structured, AI-personalized, and gamified learning environments substantially enhance cybersecurity learning outcomes, particularly for younger and less experienced users.