Effectiveness Challenges and Strategies of AI-Powered Learning Platforms for Personalised Learning

AI functions are increasingly embedded in learning management systems, adaptive tutoring, and analytics dashboards, making "AI-powered learning platforms" a practical route for delivering personalised learning at scale. Evidence from meta-analyses and selected empirical trials indicates that adaptive systems can produce small-to-moderate achievement gains, especially when recommendations respond to learners' current mastery rather than offering uniform digital practice. Yet the credibility of these gains is often contested: reported benefits are uneven across learners and settings, core algorithmic decisions remain difficult to interpret for teachers and students, and many evaluations rely on short time windows that say little about motivation, equity, retention, or transfer. Building on this literature, the paper frames effectiveness as a multidimensional construct (cognitive, behavioural, affective, and equity-related outcomes) and proposes a socio-technical account linking platform design, learner data and behaviour, teacher orchestration, and institutional governance. On this basis, it outlines actionable strategies—curricular alignment, transparent and auditable data infrastructures, structured learner support, and ethical oversight—to strengthen both the educational value and the trustworthiness of AI-enabled personalisation.