EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Personalisation and Predictive Insights: Applied AI Strategies in Modern Learning Technologies

Jarudin, Dedi

Digital Education Review · 26.06.2026 · (49) · 95-114

DOI: 10.1344/der.2026.49.95-114 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Digital Education Review, online erschienen am 26.06.2026. Verfasst von Jarudin und 1 weiteren Autor:innen. Die Arbeit nennt 42 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

When companies must grapple with ever-faster technological shifts and changing skill requirements, the classic, one-size-fits-all models of workplace learning systems fall short in meeting the needs of a diverse workforce and of predicting performance paths. We study how applied AI strategies, collaborative filtering, decision-tree classification, and neural network prediction are implemented in the three leading corporate learning technologies (Learning Management Systems—LMS, Learning Experience Platforms—LXP, and Talent Management Platforms—TMP). The primary aims include quantifying the impact of AI-enabled personalised learning on individual learner engagement and satisfaction, validating the predictive power of AI models for forecasting individual learning outcomes, and comparing the effectiveness of AI applications across different platform archetypes. We employed a mixed-methods quasi-experimental design, collaborating with three multinational organisations (technology, finance, and healthcare), with three employees from each organisation assigned to test one of the platforms (N = 450). Detailed system logs (page views, content accesses, quiz attempts), completion rates, and standardised post-training assessment scores were collected over six months, along with pre- and post-intervention survey responses (satisfaction with content, perceived relevance). Open-ended survey questions and focus group interviews provided qualitative insight into AI usability and recommendation quality. The microservice-based AI modules were integrated through platform APIs with a nightly batch-processing pipeline. Collaborative filtering applied user-item matrix factorisation to generate top-five course recommendations; decision-tree classifiers (CART) predicted the likelihood of completing a module based on early engagement metrics; and feed-forward multilayer perceptrons predicted post-assessment performance by incorporating features (demographics, prior performance, and interaction metrics). Models were trained on the first three months of data, validated in the fourth month, and deployed in the fifth and sixth months to enable real-time personalisation and prediction. Findings suggest that incorporating AI significantly enhanced learning performance. Users who received AI recommendations (n = 225) achieved 22.1 higher percentile points of active Usage compared to controls (n = 225), t(448) = 24.6, p < .001, and the proportion of passed modules increased from 62.4% to 81.7%, χ²(1, N = 450) = 48.9, p < .001. Neural network predictors achieved a mean absolute error of 0.12 on normalised assessment scores, while decision-tree models achieved an F₁-score of 0.87 for completion predictions. The top-five hit rate for collaborative filtering recommendations was 0.78. Survey satisfaction scores were higher among AI recipients (M = 4.5/5) than controls (M = 3.8/5), F(1, 448) = 182.7, p < .001. Platform-specific analyses indicated that LXPs showed a significant increase in voluntary engagement (Δ = +28.4 percentile points). TMPs provided the best support for aligning training with career development milestones (74% agree/strongly agree vs. 58% for LMS and 62% for LXP). Ethical aspects, such as informed consent, data anonymisation, and GDPR compliance, have been carefully managed. These results demonstrate that implemented AI approaches can enhance personalisation and the predictive capabilities of workplace learning technologies. The research provides a validated architecture for AI production, including modular microservices, transparent recommendation interfaces, and model retraining pipelines. It offers evidence-based suggestions to practitioners concerning platform choice and model strategy.

Zitieren

APA

Jarudin, & Dedi (2026). Personalisation and Predictive Insights: Applied AI Strategies in Modern Learning Technologies. Digital Education Review, 95-114. https://doi.org/10.1344/der.2026.49.95-114

Vancouver

Jarudin, Dedi. Personalisation and Predictive Insights: Applied AI Strategies in Modern Learning Technologies. Digital Education Review. 2026:95-114. doi:10.1344/der.2026.49.95-114

BibTeX
@article{Jarudin2026,
  author = {Jarudin and Dedi},
  title = {Personalisation and Predictive Insights: Applied AI Strategies in Modern Learning Technologies},
  journal = {Digital Education Review},
  year = {2026},
  number = {49},
  pages = {95--114},
  doi = {10.1344/der.2026.49.95-114}
}
Kostenlos starten →