From Transparency to Trust: Mitigating Perceived AI Bias to Drive Adoption

AI has advanced to the forefront of creating efficiency improvements, innovations and competitive advantages in multiple sectors. Although the application of AI into organizations is advancing rapidly, the implementation of AI in most organizations continues to face numerous obstacles including technical challenges, organizational challenges and ethical challenges. As a result of the increasing role of AI in making decisions, many individuals are concerned about the increased likelihood of issues related to bias, fairness and transparency in AI. Previous studies have examined factors affecting the adoption of new technologies which include performance expectations, amount of effort required to implement an innovation, amount of support received from upper management, preparation of the organization for the innovation, perceived risks associated with the innovation, and external pressures created by regulations or laws. To our knowledge however, very little work has been done to examine the ethics of AI, especially AI transparency and algorithmic bias. Therefore, the current research aims to understand the variables motivating managers to accept AI using two underdeveloped but important concepts - AI transparency and perceived algorithmic bias using Technology Acceptance Theory and trust theory. Data was collected from 320 individuals in IT and management fields and utilized PLS-SEM to evaluate the proposed model. Findings indicated that greater algorithmic clarity enhances user trust and reduces perceived bias, whereas perceived bias negatively impacts trust in AI. Additionally, user trust in AI serves as a mediator between both transparency and perceived bias when considering the intent to adopt. Overall, our results highlight the necessity of developing transparent and trustworthy AI systems that increase manager trust and propensity to utilize AI based tools. The research also contributes to theory and practice by providing a solid conceptual model that incorporates integrity-driven AI into the broader context of trust-based adoption.