The AI Productivity Paradox: Evidence from Firm-Level Panel Data in Pakistan

This study investigates the Productivity Paradox of Artificial Intelligence in the context of Pakistan using firm-level panel data. Despite significant investments in AI technologies, empirical evidence on productivity gains remains mixed globally. We analyze the impact of AI adoption on firm productivity, efficiency, and output across multiple sectors in Pakistan from 2015 to 2024. Using econometric models including Fixed Effects and System GMM, the results indicate a positive but delayed impact of AI on productivity, suggesting the presence of adjustment costs and complementary investments. The findings contribute to the literature on technology diffusion in developing economies and provide policy implications for accelerating AI-driven growth in Pakistan.