Bias (in ML)

A systematic tendency in data, modelling or outputs that can distort performance or produce unfair outcomes. Assess affected groups, information sources and decisions, rather than assuming overall accuracy means the system is appropriate.