International Review of Business, Trade, and Economics
The Cognitive Toll of AI: Algorithmic Over-Reliance and Macroeconomic Volatility in Institutionally Heterogeneous Emerging Economies
Abstract
Mohamed Riadh Cherif
Artificial intelligence (AI) is transforming economies, but dominant narratives treat it merely as an automation tool or productivity shock, overlooking its deeper cognitive role. This study reconceptualizes AI as Algorithmic Cognitive Augmentation (ACA)—a technology that reshapes how economic agents process information and make decisions. By bridging behavioral macroeconomics with the economics of AI, we provide the first systematic analysis of AI’s impact on macroeconomic stability through cognitive rather than purely productive channels.
We argue that the macroeconomic impact of ACA is nonlinear. Using a novel Cognitive Augmentation Index (CAI) for 20 institutionally heterogeneous emerging economies over 2005–2024, we employ Panel Smooth Transition Regression and Momentum Threshold Autoregression to identify a robust inverted U-shaped relationship. Moderate ACA is associated with higher total factor productivity, but beyond a critical threshold (CAI ≈ 0.41), over-reliance coincides with behavioral rigidities—automation bias, skill atrophy, and strategic homogenization—which are associated with a 15–25% higher macroeconomic volatility (95% CI: 12.4–27.8%).
Crucially, these patterns are strongly conditioned by institutional context: economies with weaker governance, digital infrastructure, and human capital adaptability face lower beneficial thresholds and steeper declines. Adjustment exhibits hysteresis, meaning recovery from over-reliance is slow and asymmetric.
Our findings are difficult to reconcile with the assumption that AI adoption is uniformly stabilizing. They highlight the need for context-sensitive AI governance that aligns algorithmic integration with institutional and cognitive readiness, especially in economies undergoing rapid but uneven digital transformation.