① As generative AI diffuses rapidly across the economy, expectations of a productivity revolution are rising. Yet aggregate productivity indicators have so far shown little evidence of a corresponding improvement. Using nationally representative household survey data, this paper examines whether AI adoption generates potential productivity gains through reductions in work time and whether these gains are ultimately translated into realized productivity.

② The analysis shows that AI adoption reduces average work time by 3.8 percent, equivalent to approximately 1.5 hours per week. The effect is particularly pronounced among lower-skilled workers and intensive AI users. Assuming that all time savings are reallocated to productive activities, the implied potential productivity gain is estimated at approximately 1.0 percent.

③ However, these time savings do not translate into realized productivity: the relationship between time savings and actual output growth is essentially zero. While AI has improved efficiency at the level of individual tasks, these gains have not yet extended to workflow redesign, organizational transformation, or labor reallocation, giving rise to a productivity disconnect. An important exception is found among the self-employed, professionals, and intensive AI users—groups characterized by greater job autonomy and stronger performance incentives—suggesting that organizational structure and incentive systems play a critical role in determining whether AI-generated efficiency gains are converted into higher productivity.

④ Overall, AI appears to have entered an efficiency stage but has not yet progressed to a productivity stage. This pattern is consistent with the transitional dynamics commonly associated with general-purpose technologies, including the Solow Paradox and J-curve effects. Whether AI ultimately delivers sustained productivity growth will depend on firms' ability to redesign work processes and organizational structures, reallocate tasks across workers and AI, and strengthen performance-based incentive systems. At the same time, policymakers should closely monitor how AI reshapes skill formation and career pathways, particularly for younger workers.