AI Assisted Legal Aid Accessibility and Human Support in Bangladesh

AI-assisted legal information may help people identify services and obtain preliminary guidance, but its accessibility and safety cannot be inferred from availability alone. This study examines perceptions of legal-service accessibility, human support and professional oversight using 385 adult, AI-aware survey records from Bangladesh. The sample comprises law students, graduates, advocates or legal professionals and other respondents; the workbook does not identify actual justice seekers or legal-aid users. The analysis retains individual questionnaire items and uses paired endorsement comparisons and exploratory rank correlations instead of treating heterogeneous items as validated scales. Of 385 respondents, 276 (71.7%) agreed that AI-assisted legal aid could help people identify where to seek assistance, 249 (64.7%) endorsed possible access for people far from providers, 253 (65.7%) endorsed retaining human support when AI cannot address a user’s needs, and 330 (85.7%) supported responsible AI use to complement rather than replace legal-aid professionals. The last statement exceeded endorsement of human support by 20.0 percentage points. The statements differ in content, so that difference is not a preference ranking or evidence of the relative effectiveness of safeguards. Associations of general accessibility and conditional trust in professional oversight with two accessibility items were small and uncertain after adjustment for multiple comparisons. These self-reported, cross-sectional findings do not establish actual access, service quality, improved legal outcomes or causality. The article integrates technology-acceptance and trust perspectives with people-centred access to justice and proposes future evaluation with identified legal-aid users and observed referral outcomes.