The global economy has invested over 750 billion in artificial intelligence since 2013, with global private AI investment reaching 581 billion in 2025 alone. Corporate leaders, governments, and technology vendors have unanimously promised that this investment will deliver unprecedented productivity gains. This paper argues that those promises are systematically overstated and, in most cases, empirically unverifiable. Drawing on longitudinal productivity data from the IMF, OECD, Bureau of Labor Statistics and World Bank, supplemented by systematic document analysis of major institutional AI surveys spanning 2023 to 2026, this study presents four empirical findings that challenge the dominant productivity narrative at a global scale. First, AI investment and total factor productivity are moving in opposite directions across every major economy simultaneously. Second, experienced workers using AI tools take 19% longer to complete tasks than those working without them yet believe they are 20% faster. Third, 56% of global chief executives report getting nothing from their AI investments yet continue to accelerate spending. Fourth, genuine AI productivity gains are confined to just 5% of enterprise deployments. This paper introduces the Productivity Illusion Framework to explain the three mechanisms sustaining this gap and proposes the Measurement Trap as the self-reinforcing process through which organisations continue investing despite evidence of underperformance. The paper also introduces the Workslop Economy as a new theoretical construct describing the emergent category of AI-generated organisational waste that is consuming the productivity gains AI nominally creates.
Pranjal Sameliya (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: