Artificial Intelligence (AI) has become an integral component of decision-making processes in contemporary society, influencing outcomes in finance, healthcare, education, employment, criminal justice, and public administration. While AI systems are frequently described as objective, data-driven, and efficient, a growing body of research indicates that they can reproduce or even intensify existing social biases. These biases often stem from imbalanced datasets, flawed model design, and a lack of transparency and accountability. This paper critically examines the issue of bias and fairness in AI-driven decision systems, with particular emphasis on opaque “black-box” models used in high-stakes contexts. The study explores key sources of algorithmic bias, reviews foundational fairness concepts such as individual and group fairness, and discusses commonly used fairness metrics. A mixed-method research approach combining literature analysis and survey-based data collection is employed to assess public perceptions of AI fairness. The findings reveal a significant trust deficit between users and automated systems and highlight the importance of Explainable Artificial Intelligence (XAI), human-in-the-loop mechanisms, and governance frameworks. The paper concludes that achieving fair and accountable AI requires a holistic socio-technical approach that integrates technical, ethical, and institutional interventions.
Ayesha et al. (Sat,) studied this question.