Artificial intelligence adoption: productivity effects and supporting policies

In Italy, the share of firms adopting artificial intelligence technologies is growing but remains below the European average. Long-run estimates suggest that the diffusion of this technology could raise productivity by 0.2 - 1.1 percentage points per year over the next decade, depending on the speed and depth of adoption. In the short run, however, no significant effects on firm-level productivity are apparent yet, despite documented gains at the level of individual tasks: this outcome is consistent with the literature, in which aggregate benefits only materialize after substantial organizational adjustments. The paper identifies the main barriers to adoption and the conditions that justify public intervention. Drawing on an analysis of the policies adopted in more advanced countries, it highlights the need for a coherent strategy that, rather than relying on poorly targeted subsidies, supports firms through the adoption process, fosters the supply of applications by encouraging the development of specialized providers, and invests in enabling factors: regulatory certainty, data sharing, and ease of access to computing infrastructure.