Understanding dark side of artificial intelligence (AI) integrated business analytics: assessing firm’s operational inefficiency and competitiveness

Journal: European Journal of Information Systems

Date: 2022

Author: Rana, N.P., Chatterjee, S., Dwivedi, Y.K., and Akter, S.

Abstract:
The data-centric revolution generally celebrates the proliferation of business analytics and AI in exploiting firm’s potential and success. However, there is a lack of research on how the unintended consequences of AI integrated business analytics (AI-BA) influence a firm’s overall competitive advantage. In this backdrop, this study aims to identify how factors, such as AI-BA opacity, suboptimal business decisions and perceived risk are responsible for a firm’s operational inefficiency and competitive disadvantage. Drawing on the resource-based view, dynamic capability view, and contingency theory, the proposed research model captures the components and effects of an AI-BA opacity on a firm’s risk environment and negative performance. The data were gathered from 355 operational, mid-level and senior managers from various service sectors across all different size organisations in India. The results indicated that lack of governance, poor data quality, and inefficient training of key employees led to an AI-BA opacity. It then triggers suboptimal business decisions and higher perceived risk resulting in operational inefficiency. The findings show that operational inefficiency significantly contributes to negative sales growth and employees’ dissatisfaction, which result in a competitive disadvantage for a firm. The findings also highlight the significant moderating effect of contingency plan in the nomological chain.

Link: Google Scholar

Background and Context

Research Focus

This study investigates the negative consequences of AI integrated business analytics (AI-BA) on firm's operational efficiency and competitive advantage when implemented inappropriately.

Theoretical Foundation

The research draws on resource-based view, dynamic capability view, and contingency theory to develop and test a model of AI-BA opacity effects.

Methodology

Data were collected from 355 operational, mid-level, and senior managers across various service sectors in India and analyzed using structural equation modeling.

Moderating Effect of Contingency Planning on Operational Inefficiency

A strong contingency plan significantly reduces the negative impact of suboptimal decisions on operational inefficiency.
The moderating effect of contingency planning is statistically significant (p < 0.05) with coefficients of 0.17 and 0.23.
This finding suggests that firms can mitigate AI-BA risks through proper contingency planning.

Full Structural Model Explaining Competitive Disadvantage

The full structural model explains 67% of competitive disadvantage variance, demonstrating high predictive power.
All hypothesized relationships were supported, confirming the negative impact flow from AI-BA opacity to competitive disadvantage.
The model highlights critical intervention points where firms can break the chain of negative effects.

Contribution and Implications

The study offers first empirical evidence on how AI-BA opacity negatively impacts firm performance through multiple pathways.
Findings extend the emerging "explainable AI" body of research by identifying components of AI-BA opacity.
Results provide guidelines for managers to develop robust AI governance, data quality standards, and employee training programs.
The research highlights contingency planning as a critical intervention to mitigate AI risks and operational inefficiency.
The study emphasizes the importance of transparent, explainable AI systems for maintaining competitive advantage.

Data Sources

The path coefficients visualization is based on data from Table 5 showing structural model results and path coefficients.
The operational impact chart uses data from Table 5 showing R² values and path coefficients between constructs.
The full model visualization is based on the validated research model shown in Figure 4 of the article.
The AI-BA opacity components visualization is based on Tables 2 and 3 showing measurement properties and second-order relationships.
The contingency planning visualization is derived from Table 7 and the moderation analysis in Figures 5 and 6.