When does AI capability improve sustainable performance? Strategic agility and the dark side of top management support
Emmanuel Nketiah, Mavis Adjei, Caohong WangPurpose
Returns on artificial intelligence (AI) capability vary widely. Most research links AI investment directly to performance without explaining what converts capability into results. This study positions strategic agility as that conversion process and tests how readiness, executive support and culture shape it.
Design/methodology/approach
The study integrates dynamic capabilities and sociotechnical theory into a process-based contingency model. It applies partial least squares structural equation modeling (PLS-SEM) to two-wave survey data from 522 Fujian firms. Multigroup analyses test how effects differ by firm size and sector.
Findings
Strategic agility partially mediates the AI-sustainability link, and employee AI readiness strengthens this mediation. Top management support weakens it instead, a dark-side pattern tied to institutional decoupling. Green culture shows no moderating effect. AI capability harms small firms but benefits larger ones; the direct pathway holds only in industrial sectors.
Practical implications
AI creates sustainability value when firms treat it as a capability, not a self-executing asset. That capability needs agile routines and employee readiness to turn AI-generated insight into action. Distributed experimentation also merits protection: top-down endorsement can erode the innovation it aims to promote.
Originality/value
Prior research has largely linked AI investment directly to performance, or treated sensing, seizing and reconfiguring as parallel capabilities. This study specifies strategic agility as the mechanism that converts AI capability into sustainable performance, moves beyond input-output logic and identifies a counterintuitive dark side of top management support, explained by institutional decoupling. For managers, AI's sustainability returns depend on firm scale and sector, not universality. For policymakers, readiness-building and governance support best improve AI's payoffs.