What is the J-curve of AI productivity, and what does it mean for organisations investing now?

The J-curve of AI productivity refers to a pattern observed in transformative technologies, where initial adoption leads to a temporary dip in measured productivity due to upfront costs in learning, experimentation, and intangible investments like skills and knowledge required to integrate AI effectively [1][4][5]. This "learning period" delays visible gains as organizations invest in R&D, education, and infrastructure before the benefits materialize, potentially causing mismeasurement in economic data [5][7]. For AI specifically, this curve suggests early challenges in aggregating micro-level task improvements into firm-wide productivity, despite sector-specific enhancements [7][10]. For organizations investing now, the J-curve implies that bold, early experiments—often low-cost—can yield outsized returns as AI adoption accelerates exponentially, while a "fast follower" strategy risks falling behind in a rapidly evolving landscape [4]. Surveys of executives indicate positive labor productivity gains from AI, expected to strengthen in 2026, though with heterogeneity across sectors and little near-term employment decline [8]. Investing in public goods like infrastructure positions firms to capture long-term efficiency and value creation once the surge phase begins [2][9].