Designing Inclusive AI for Aging Populations: A Conceptual Agentic Framework for Lifelong Learning

Global demographic changes are also changing the educational focus, where the aged will number over two billion by 2050. Despite progress in innovative artificial intelligence (AI) in education, most frameworks are developed for younger learners, who are underserved. This paper introduces the Agentic AI framework for elderly learners, a conceptual model which makes seniors active, autonomous learners throughout their life. The framework entails the combination of three mutually supporting layers, i.e., emotion-sensitive mentorship, where learning interactions are personalized to emotional states; gamified discovery learning, where learning engages curiosity and maintains motivation; or reflective knowledge reservoir, where knowledge is anchored by memory and cognitive resilience. A systematic review of recent literature suggests that although each of these domains has been independently validated, there is no common framework to address integration of the domains in elderly learners. The paper formulates three research questions and corresponding hypotheses and calls them onto the proposed layers and a roadmap for research. This roadmap progresses to consolidation on the theoretical level to prototype development, pilot testing, and large-scale validation to ensure that it is scalable and inclusive. The discussion not only introduces some theoretical contributions such as the extrapolation of the affective computing, gamification and reflection to the field of elderly learning, but also introduces some practical implications such as design solutions such as empathetic AI mentor, age-friendly gamification and age-friendly reflection. There is also focus on societal effects, which help to indicate how the framework can contribute to cognitive health, social inclusion and fair digital participation. Limitations such as the lack of empirical testing is noted along with directions of future research. The combination of discontinuous strands into a unified model provided through the Agentic AI framework provide a stringent basis to create inclusive learner-centered AI systems that will allow the elderly adults to excel in the digital era.