Rogue AI and what to do about racist algorithms

What happens when machine learning algorithms pick up on human biases? Naturally, the bias scales.
This article covers the extraordinary challenge that machine learning practitioners face when they set their algorithms loose on training datasets. What will be the consequences of their data choice? In what context will the results be used?
The results can often be surprisingly offensive – and the anti-bias techniques now being employed to counter this are becoming more and more sophisticated.