The paper proposes the AI Journal, a persistent, user‑controlled personal AI environment designed to serve as a lifelong developmental scaffold rather than a replacement for human relationships or professional judgement. Drawing on self‑determination theory, attachment theory, sociocultural learning, and ecological‑developmental perspectives, the author argues that persistence, personalisation, and user control can support autonomy, competence and relatedness if bounded by strict psychological and data‑sovereignty safeguards. Five core functions are outlined – reflective dialogue, adaptive learning, longitudinal self‑understanding, practical planning, and selective community matching – each predicated on privacy‑by‑default, contestable memory, purpose limitation, anti‑dependency design, age‑appropriate safeguards, and human‑led escalation for safeguarding. The paper critically examines risks (surveillance, manipulation, dependency, identity fossilisation, institutional capture) and proposes a narrow school‑based prototype, MyVoice, as a daily check‑in to test feasibility, acceptability and early detection of adverse experiences such as bullying. Ten testable propositions (P1‑P7) are presented, forming a research programme that emphasizes longitudinal outcomes, autonomy‑supportive design, and rigorous safeguarding. The work situates the AI Journal within a broader interdisciplinary One Earth programme, emphasizing that its value depends on institutional design, not on technological determinism, and calls for empirical longitudinal studies to evaluate whether persistent, reflective AI support can improve developmental trajectories while preserving data sovereignty and human flourishing.
Topics
Reframing AI in Education as a Persistent Developmental Scaffold
Positions AI not as a tool or chatbot but as a longitudinal personal record and reasoning environment that can accompany a learner across life stages.