• Since the adoption of generative AI, working hours have decreased, but the expected benefits have not translated into improved organizational performance, resulting in productivity leakage.

• This mirrors the historical “productivity paradox” observed during the early adoption of computers, where the impact of new technology was offset by factors such as redundant tasks, underutilization, and inadequate role realignment.

• Productivity leakage is particularly evident when AI is simply overlaid onto existing workflows, or when organizations lack structured plans for utilization, performance measurement systems, and cultural readiness.

• To address this, change management must be built into the AI implementation process. As seen in the case of DBS Bank in Singapore, combining AI adoption with job redesign and redeployment significantly amplifies productivity gains.

• Ultimately, moving beyond a technology department-centric approach and establishing company-wide role allocation and coordination across strategy, finance, HR, and other functions is key to enhancing AI productivity.