Bias dichotomy in the age of business AI copilots: a systematic literature review

Despite the increased efficiency of Artificial Intelligence (AI), it introduces ethical challenges and biases. AI copilots - virtual assistants that support managers in decision-making - exemplify these issues by embedding biases from algorithms that reflect cultural norms, thereby limiting free expression on sensitive topics. This research aims to systematically analyse biases in AI copilots within the field of management, distinguishing between data-driven and algorithmic biases, and developing mitigation strategies to enhance understanding of bias dynamics in AI systems. Employing a qualitative methodology through a systematic literature review, the study extrapolates data from 575 selected documents and organises the key findings. It uncovers biases present in every phase of the data and AI system lifecycle, along with the underlying reasons for their persistence. This comprehensive understanding contributes to insights into how biases manifest and impact managerial decision-making processes, ultimately enriching the discourse on organisational governance.