AI-embodied agricultural technologies, population ageing, and grain production: evidence from China

China has rapidly expanded the use of plant-protection drones (UAVs) amid population ageing and growing concerns about sustaining grain production. Using a panel of 31 provincial-level regions over 2018–2024 and two-way fixed-effects models, we examine whether the association between UAV diffusion and grain output depends on population ageing, and whether this dependence differs between AI-embodied and conventional agricultural machinery. The association between UAV stocks and grain output weakens as the share of the population aged 60 and above rises. The result is robust to lagged regressors, additional controls for disaster exposure and farm structure, the exclusion of Tibet, differential year effects for the main grain-producing region, and alternative measures of ageing, but not to province-specific linear trends. A decomposition indicates that the negative interaction is concentrated in grain sown area rather than yield. When UAV and large-tractor stocks are estimated jointly, the ageing interaction is significantly more negative for UAVs than for tractors. The estimates should be interpreted as conditional associations rather than causal effects. The findings are consistent with the possibility that AI-embodied agricultural technologies face greater demographic constraints than conventional machinery and suggest that machinery policies may need to account for differences across technology types.