Authors: Navya Sri Maddukuri, Hari Nagakoteswar Tripurari
Abstract: As artificial intelligence systems become embedded in core business processes, organizations increasingly require governance mechanisms that translate high-level ethical principles into operational capabilities capable of preventing, detecting, and recovering from AI-related failures. This study develops and validates an AI Governance Maturity Index (AGMI) spanning six dimensions — policy formalization, risk monitoring, model documentation, human oversight, incident response, and continuous auditing — and examines its relationship to organizational digital resilience. Drawing on survey data, archival AI incident records, and six in-depth case studies from 341 firms deploying high-impact AI systems across six industry sectors, the study tests whether governance maturity reduces AI failure frequency and severity and improves post-incident recovery. Regression results indicate that AGMI is significantly associated with lower incident frequency (β = –0.58, p < .001), lower incident severity (β = –0.49, p < .001), shorter mean time to recovery (β = –8.92 hours, p < .001), and higher Digital Resilience Index scores (β = 0.11, p < .001), with continuous auditing and human oversight intensity emerging as the dimensions most strongly associated with resilience outcomes and exhibiting significant complementary interaction effects. A five-tier maturity comparison reveals a tenfold difference in mean time to recovery between Tier 1 (Ad Hoc) and Tier 5 (Adaptive) firms (78.4 versus 9.8 hours). A twelve-month governance investment pilot across 48 firms demonstrates that combined investment in continuous auditing, cross-functional incident response, standardized documentation, tiered oversight, and automated risk telemetry produces a 1.54-point AGMI gain and a 3.67-incident annual reduction, substantially exceeding the effect of any single investment. Six case studies of firms experiencing significant AI incidents illustrate the governance-resilience feedback loop through which incident response activates targeted maturity advancement. The paper contributes a multi-level (macro-meso-micro-outcome) model linking responsible AI practices to digital resilience, a validated maturity instrument, and a practical roadmap for firms seeking to scale AI deployment while maintaining trust, regulatory compliance, and organizational accountability.
DOI: http://doi.org/10.5281/zenodo.20837668
