Digital dependence can support educational continuity during disasters and other major disruptions, but it can also magnify existing inequalities when access and institutional capacity are uneven.
This study assessed digital risk and institutional preparedness for artificial intelligence (AI)-enabled education in Sherpur Sadar Upazila, Bangladesh. A convergent mixed-methods design combined a 14-item survey of 207 students from five urban and three rural educational institutions with qualitative evidence from 25 education stakeholders. Survey data were analyzed descriptively, with nonresponse treated as missing; qualitative data were examined thematically and integrated through a joint display.
Students strongly supported government investment in rural digital access (98.1%) and greater institutional use of technology (74.9%). At the same time, 72.5% reported that technology was difficult to use, 66.7% identified financial barriers, and more than 84% reported social or family constraints.
Qualitative findings identified interconnected vulnerabilities involving infrastructure deficits, gendered digital exclusion, limited teacher preparation, weak technical and institutional support, socio-cultural restrictions, and policy-implementation gaps.
The integrated findings reveal a preparedness paradox: favorable attitudes toward technology coexist with conditions that can undermine equitable participation and continuity during floods, cyclones, public-health emergencies, infrastructure failures, and other disruptions.
Educational resilience therefore requires reliable connectivity and electricity, affordable devices, offline and backup learning arrangements, teacher capacity development, gender-responsive access measures, accountable AI governance, and the integration of digital continuity into institutional disaster-preparedness planning.
The study contributes a risk-informed interpretation of technology acceptance while avoiding claims that the instrument constitutes a full validation of TAM3 or a direct measure of actual AI use.
Authors: Ritesh Karmaker, Vladimir M. Cvetković, Vanja Cvetković, Hatidža Beriša, and Srđan Milašinović
