Do AI policy dividends alleviate employment pressure on college graduates? A comparative study across china provinces

China's rapid expansion of higher education has intensified graduate employment pressures under uncertain labor market conditions. This study examines whether provincial artificial intelligence policies alleviate these pressures, measured through a composite index capturing graduate unemployment rates, job placement ratios, and time to first employment. Using a balanced panel of 31 provinces from 2016 to 2023, we implement a multi-method identification strategy combining two-way fixed effects, staggered difference-in-differences with event-study validation, instrumental variables estimation using lagged national AI funding and historical patent densities, and a Spatial Durbin Model to capture cross-provincial spillovers. Results indicate that AI policy adoption reduces the employment pressure index by 3.7 to 6.4 percent, with event-study plots confirming flat pre-trends and instrument diagnostics satisfying relevance and exogeneity requirements. Heterogeneity analysis reveals substantially stronger effects in provinces with higher digital literacy, greater firm-level AI adoption, and stronger policy implementation capacity, explaining much of the observed East–non-East differential. Spatial estimates identify significant positive spillovers operating through labor mobility, supply chain linkages, and knowledge diffusion channels. These magnitudes translate to several thousand additional graduate placements annually per adopting province. The findings suggest that regionally coordinated AI strategies, coupled with targeted capacity-building in lagging provinces, can meaningfully ease structural employment challenges facing Chinese graduates.