Contestable AI by Design: Towards a Framework

Abstract As the use of AI systems continues to increase, so do concerns over their lack fairness, legitimacy and accountability. Such harmful automated decision-making can be guarded against by ensuring are contestable design: responsive human intervention throughout system lifecycle. Contestable design is a small but growing field research. However, most available knowledge requires significant amount translation applicable in practice. A proven way conveying intermediate-level, generative form frameworks. In this article we qualitative-interpretative methods visual mapping techniques extract from literature sociotechnical features practices that contribute AI, synthesize these into framework.