AI-Supported Lifelong Learning: Predicting Digital Competence and Learner Profiles with PIAAC Turkey Data

Purpose: Although AI offers enormous potential for facilitating adaptive learning for diverse learners at scale, it could disproportionately neglect older cohorts and other digitally marginalized populations, with most benefits accruing to younger generations instead. In this paper, we examine how analytics can utilize AI to model digital literacy and learning requirements using PIAAC microdata from Turkey. Methods: Feature sets with diverse variables were constructed, including indicators for technology-based problem solving, plausible values for literacy and numeracy, indices related to technology utilization and accessibility, demographic factors, and engagement and motivation indicators. After weight-aware preprocessing, supervised machine learning models—Random Forest classifier, Gradient Boosting classifier, Support Vector Machine classifier, and Multi-Layer Perceptron classifier—were applied to predict continuous digital competence scores and to classify high/low digital competence. SHAP analysis was used to interpret the machine learning models. Learner profiles were established using UMAP and k-means clustering techniques. Results: Gradient Boosting and Random Forest models demonstrate the best compromise between discrimination power and calibration of predicted probabilities. Literacy, numeracy, engagement with digital technologies, and motivational variables emerge as dominant factors, while a strong negative age gradient suggests intergenerational disparities. Three distinct learner types—low-, moderate-, and high-level learners—differ significantly across clusters, defining distinct patterns of learning needs. Conclusion: The findings suggest that AI-supported lifelong learning systems need to link skill-building with motivational components and continuous fairness 1tracking. Human-centered AI design is required to avoid compounding inequalities and to guarantee equitable access to innovative digital learning tools for older adults and other at-risk populations.