AI Leteracy In Social Work Education: Student Perspectives on Use, Benefits, and Barriers

Background and Purpose: Artificial Intelligence tools are now a routine feature of higher education, yet social work programs have been slow to respond. How students are actually adopting these tools and what drives or suppresses that adoption remains poorly understood. Understanding this technology and applying it ethically is not only imperative for curriculum design but for equity. When programs are silent on AI, students with greater prior access and technical confidence are advantaged, and those without are left behind. Grounded in the Technology Acceptance Model, Diffusion of Innovations, and Social Cognitive Theory, this study examines what predicts AI adoption among social work students across a 2×2 outcome framework current use versus future intentions, in coursework versus field practice. This study considers what the findings mean for a profession built on democratic values and service to vulnerable populations.

Methods: A cross-sectional survey, delivered by email, was completed by 150 social work students. Quantitative measures included AI knowledge and self-efficacy, perceived benefits, perceived barriers, attitudinal orientations, and frequency of AI use across academics and practice. Four multiple regression models tested predictors of each outcome in the 2×2 framework. Supplementary analyses examined item-level benefit and barrier priorities, perceived peer norms, and subgroup differences by age and field placement status. The study was IRB-approved through CUNY.

Results: Students rated barriers considerably higher than benefits (M = 3.69 vs. 2.38) yet barriers did not significantly predict use in any of the four regression models once benefits, knowledge, and attitudes were controlled. Current use in coursework (R² = .520) and practicum (R² = .399) were predicted by the same triad: AI knowledge and self-efficacy, perceived benefits, and pro-AI attitudes. Future intentions of AI use told a different story. Planning to use AI in coursework was driven almost entirely by pro-AI attitudes (β = .532), while professional practice intentions, the strongest model (R² = .592), required the combination of attitudes, perceived benefits, and competence. Students believed more of their peers used AI than they themselves reported (t(144) = 5.72, p < .001). The 21% who reported never used AI showed large effect-size differences across every predictor domain (d = 0.77–1.77), This suggests a non-adopter group that is qualitatively distinct, not simply at the low end of a continuum.

Conclusions and Implications: Social Work students are adopting AI individually with little discussion with peers, for practical reasons and without programmatic guidance. Concerns about using AI are real but don't stop use. The absence of institutional effects is not a neutral finding; it reflects a gap in professional education at a moment when the stakes are high. Preparing graduates to use AI ethically and equitably in ways that serve rather than harm the people social work exists to support requires programs to move beyond policy statements and into the curriculum itself.