From Andy Biddulph
Douglas Heaven inadvertently exposes the real danger of this type of AI, anthropomorphising complex algorithms. Machine learning, having no moral component, cannot be racist, as was claimed when Google’s search function ended up directing ads asking “have you ever been arrested?” towards black people.
What it did find was a correlation between having a black person’s name and involvement with the criminal justice system.
There are no profound ethical dilemmas if we accept AI for what it is, an observer of statistical correlations. I suppose it is more comfortable to call a box of transistors racist than face up to our own prejudices.
Burton upon Trent,Staffordshire, UK
