Аңдатпа
This paper develops and validates a methodological framework for an AI-based distance learning platform grounded in core AI principles (data-driven decision-making, learner modeling, adaptive personalization, and continuous feedback). The study aims to (i) formulate a structured development methodology for an AI-enabled platform and (ii) provide empirical evidence of its effectiveness using measurable learning and engagement indicators. The proposed structural–methodological model integrates four interconnected components: (1) platform architecture (data acquisition, content delivery, assessment services, analytics & AI layer, administration/security); (2) personalization mechanisms (learner model, adaptive learning trajectories, recommendation engine, analytics-driven feedback); (3) effectiveness evaluation metrics (pre/post test scores, learning gain, engagement indicators, learner satisfaction); and (4) normative and pedagogical requirements (data protection, academic integrity, accessibility, alignment with intended learning outcomes). The framework was piloted through a quasi-experimental study conducted at Kh. Dosmukhamedov Atyrau University, Department of Computer Science, within the course “Artificial Intelligence Platforms,” involving 2nd-year MSc students (N=22; Control n=11, Experimental n=11). Descriptive results showed that the Control group improved from 57.56% (pre-test) to 69.55% (post-test), yielding a gain of 11.99 percentage points (pp) and a 20.86% relative gain. In contrast, the Experimental group improved from 51.05% to 73.30%, yielding a gain of 22.25 pp and a 44.08% relative gain. The effectiveness of the AI-based platform was statistically supported: Welch’s t-test on gain scores indicated a significant group difference (t=5.397, df=20.0, p<0.001), with a very large effect size (Cohen’s d=2.30) and a 95% CI for the mean gain difference of Δ=[6.30, 14.23] pp. A baseline-adjusted ANCOVA (post-test as the dependent variable, group as the factor, pre-test as the covariate) confirmed a significant group effect (F=26.323, df1=1, df2=19, p=0.0001) with substantial explained variance (partial η²=0.581). Overall, the findings demonstrate that AI-driven personalization can substantially increase learning gains and strengthen the stability of outcomes in distance learning environments.