Can AI Empower Financial Social Work? Evidence from a Quasi-Experimental Study of AI-Enhanced Financial Education In China

Background and Purpose: Leveraging artificial intelligence to enhance financial education has emerged as a critical priority, yet evidence on the effectiveness of AI-powered financial literacy interventions remains limited, particularly among adolescents in developing country contexts. This study evaluates the "Star Sea AI Academy" — a collaborative initiative between the China Financial Education Development Foundation and Ant Group — which integrated AI-assisted tools into school-based financial literacy curricula for adolescents across China. The primary purpose is to assess whether AI-empowered financial education produces stronger financial literacy outcomes than traditional instructional approaches, and to identify the specific competency domains where AI instruction demonstrates comparative advantages.

Methods: Using a quasi-experimental design with non-randomized assignment and pre-test/post-test measurement, students were assigned to either an AI-empowered instructional group or a traditional instruction comparison group. Financial literacy was assessed at baseline and following the curricular intervention using structured questionnaires. Baseline comparability between groups was confirmed prior to analysis (AI group: M=18.99; basic group: M=18.51). Quantitative data were supplemented by interviews with teachers and students conducted between testing periods. Cross-sectional between-group comparisons and longitudinal within-group tracking were conducted to assess intervention effects.

Results: Both groups demonstrated significant improvements in financial literacy following the intervention. The traditional instruction group showed a larger gain on the primary financial literacy measure (M=18.51 to M=21.99, +3.48 points) compared to the AI group (M=18.99 to M=20.59, +1.60 points). Paired-samples t-tests confirmed that both within-group improvements were statistically significant: the AI group exhibited a significant increase (t=-2.48, p=0.01), improving in core financial literacy, financial knowledge, financial attitude, financial behavior, and all three AI-specific competencies (all p<0.05). The traditional group showed a highly significant improvement (t=-5.08, p<0.001), with large effects on core financial literacy (d=0.77), financial knowledge (d=0.909), and financial behavior (d=0.976). ANCOVA results confirmed a significant between-group difference after controlling for pre-test performance (β=-1.698, p=0.037), indicating the traditional group's gain was statistically larger. However, effect-size Z-tests showed no significant differences across core domains (p>0.05). The two instructional models produced distinct competency profiles: the AI group showed moderate effects on financial knowledge, financial behavior, and AI-assisted practice competence (d=0.631–0.677), and demonstrated unique advantages in higher-order competencies — specifically the transformation from financial knowledge to applied financial capability — and in AI-specific financial practice skills not captured by the primary literacy measure. Both groups reported high satisfaction ratings.

Conclusions and Implications: AI-empowered financial education shows meaningful potential for advancing financial social work practice, particularly in developing higher-order financial capability among adolescents. The finding that traditional instruction produced larger gains on foundational literacy warrants careful interpretation — AI-enhanced approaches may require curricular refinement and stronger teacher capacity to maximize effectiveness across all domains. These findings suggest AI tools are most valuable as a complement to, rather than replacement for, evidence-based financial education. Equitable access requires attention to teacher training, tiered curriculum design, and school-family collaboration. Future research should employ randomized designs and longer follow-up periods to establish causal effects.