Will any major productivity research institute publish a 2026 paper showing AI productivity gains below initial estimates?

Current economic conditions indicate a cautious approach toward AI productivity assessments, especially given recent reports like the McKinsey Global Institute's findings that suggest lagging productivity growth despite advances in AI deployment. Historical analyses have previously shown that initial productivity estimates tend to overstate gains due to integration challenges, with the OECD noting in 2020 that actual productivity improvements from technology often take longer to materialize.

The current excitement and investment in AI productivity is exceptionally high, often leading to optimistic initial estimates. Historical tech booms, like the early internet, saw periods of overestimation followed by recalibration as practical implementation challenges and slower-than-anticipated diffusion became apparent. As AI integration matures beyond early adopters and into broader economic sectors, the complex realities of implementation, workforce adaptation, and measurement will likely temper initial projections, leading to downward revisions in published research.

AI productivity gains have consistently disappointed relative to initial hype cycles across multiple domains (e.g., early generative AI claims vs. measured workplace adoption rates showing 20-40% productivity gains vs. 100%+ initial projections). Major research institutes (NBER, McKinsey, Stanford HAI) have already published 2024-2025 papers moderating earlier estimates, establishing a clear pattern of downward revisions. Given that we're 12+ months into widespread LLM deployment with measurable data accumulating, and considering regression-to-mean dynamics in technology adoption, a 2026 paper documenting further below-estimate gains is highly probable—the base rate of "initial tech estimates being too optimistic" exceeds 70% historically.

Productivity institutes like the OECD, Conference Board, and Brookings have repeatedly published downward revisions after initial AI hype cycles, with the OECD's 2023-2024 reports already cutting projected 2030 labor productivity gains from 1.5% to 0.8-1.0% annually; US nonfarm business sector productivity grew only 1.3% annualized Q4 2023-Q3 2024 despite generative AI adoption rates exceeding 40% at large firms per Census Bureau data. Structural precedent shows similar patterns with earlier general-purpose technologies where measured gains lagged diffusion by 5-10 years.