Designing ethical AI: balancing inclusion and autonomy

Purpose This paper aims to explore the ethical tension between representational inclusivity and individual freedom in AI systems. It highlights how bias mitigation efforts, especially for women and children, may unintentionally restrict user autonomy when ethical interventions prescribe engagement patterns. Design/methodology/approach Using a data-centric and human-centered design, the study integrates the FAT (Fairness, Accountability, Transparency) model and Stanford’s Human-Centered AI framework. It analyses systemic nudging through empirical illustrations from recommendation systems, education platforms and profiling tools. Findings Ethical interventions can compromise autonomy if not carefully designed. A dual-layered framework – combining system-level bias auditing with user-level content mediation – is proposed to balance fairness with freedom. Research limitations/implications The study relies on illustrative cases and literature. Future work should include empirical validation through original data collection and longitudinal analysis. Practical implications The framework offers guidance for developers, educators and policymakers to design AI systems that balance personalization with ethical representation. Social implications The study promotes inclusive representation and user empowerment, contributing to equity and justice in AI technologies. Originality/value By integrating ethical theory with practical illustrations, the paper presents a structured approach to designing AI systems that respect both representation and autonomy.