Adolescent hikikomori-like social withdrawal is a form of contemporary vulnerability that escapes traditional clinical-diagnostic tools, as it lies in a grey zone between physiological withdrawal and pathological isolation. Recent scientific evidence reports a marked increase in implicit dropout and prolonged self-isolation among Italian adolescents. As the primary educational agency, schools are called to rethink their tools for the early detection of invisible fragilities. This contribution develops a critical-pedagogical proposal for integrating AI-based Early Warning Systems (EWS) and Learning Analytics into secondary education, orienting them toward the early detection of hikikomori-like withdrawal trajectories. The proposal critically engages with the risks documented by scholarship on educational data, namely algorithmic surveillance, predictive labelling, epistemic injustice, and cognitive delegation (Prinsloo & Slade 2016b), treating them as constraints on pedagogical legitimacy. The study first develops a narrative review of empirical and design-oriented literature on three nuclei, namely Early Warning Systems, machine learning-based dropout prediction, and adolescent hikikomori, and then proceeds with a documentary analysis of the main supranational and national policy frameworks on AI in education (UNESCO 2023; European Union 2024; MIM 2025). Drawing on Saito (2026) and on the tripartition formalised in Italian school policy of teaching with, about, and for AI, Open Journal of Humanities, 22 (2026) issn 2612-6966 500 four operational conditions of legitimacy are proposed, namely teacher-inthe- loop mediation as a safeguard against opacity and cognitive delegation, transparency and traceability of algorithmic risk attributions, AI literacy oriented toward critical awareness and learner autonomy, and a capabilityand care-driven design that recognises hikikomori students as epistemic subjects. The study reframes the debate on AI-mediated early detection within a pedagogically and ethically grounded paradigm, reading hikikomori-like withdrawal as a contemporary expression of invisible fragility and outlining capacitating learning ecologies anchored in special and inclusive pedagogy.
Sistemi di allerta precoce basati sull’IA e hikikomori nella scuola secondaria
Verbicaro M.;Oliva P.
2026-01-01
Abstract
Adolescent hikikomori-like social withdrawal is a form of contemporary vulnerability that escapes traditional clinical-diagnostic tools, as it lies in a grey zone between physiological withdrawal and pathological isolation. Recent scientific evidence reports a marked increase in implicit dropout and prolonged self-isolation among Italian adolescents. As the primary educational agency, schools are called to rethink their tools for the early detection of invisible fragilities. This contribution develops a critical-pedagogical proposal for integrating AI-based Early Warning Systems (EWS) and Learning Analytics into secondary education, orienting them toward the early detection of hikikomori-like withdrawal trajectories. The proposal critically engages with the risks documented by scholarship on educational data, namely algorithmic surveillance, predictive labelling, epistemic injustice, and cognitive delegation (Prinsloo & Slade 2016b), treating them as constraints on pedagogical legitimacy. The study first develops a narrative review of empirical and design-oriented literature on three nuclei, namely Early Warning Systems, machine learning-based dropout prediction, and adolescent hikikomori, and then proceeds with a documentary analysis of the main supranational and national policy frameworks on AI in education (UNESCO 2023; European Union 2024; MIM 2025). Drawing on Saito (2026) and on the tripartition formalised in Italian school policy of teaching with, about, and for AI, Open Journal of Humanities, 22 (2026) issn 2612-6966 500 four operational conditions of legitimacy are proposed, namely teacher-inthe- loop mediation as a safeguard against opacity and cognitive delegation, transparency and traceability of algorithmic risk attributions, AI literacy oriented toward critical awareness and learner autonomy, and a capabilityand care-driven design that recognises hikikomori students as epistemic subjects. The study reframes the debate on AI-mediated early detection within a pedagogically and ethically grounded paradigm, reading hikikomori-like withdrawal as a contemporary expression of invisible fragility and outlining capacitating learning ecologies anchored in special and inclusive pedagogy.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
