GDP and the Limits of Macroeconomic Measurement in the Age of Artificial Intelligence

Purpose - This article examines how the diffusion of artificial intelligence is widening a long-running gap between what GDP records and what economies now produce, and what this implies for the macroeconomic measurement framework on which policy still depends. Design/methodology/approach - The argument is conceptual. It rests on a synthesis of three established literatures and a reading of recent empirical evidence on AI diffusion through that lens. Findings - AI enlarges this gap through two channels. The first is the spread of free digital goods that users value highly but rarely pay for, with unrecorded consumer surplus from generative AI now estimated in the tens of billions of dollars annually. The second is the distance between the heavy investment now flowing into AI and the modest productivity the statistics so far record, a lag familiar from earlier general-purpose technologies. Originality/value - The article develops an integrated framework that connects three literatures usually examined separately: the welfare critique of GDP, data economics, and the diffusion of general-purpose technologies. Through this framework, artificial intelligence emerges as a contemporary stress test of macroeconomic measurement, revealing a growing divergence between GDP and the value the economy actually produces.