Why it matters
There are many subtle ways in which a seemingly well-intentioned and neutral algorithm can yield unintended biases. These mistakes are relatively benign if, for example, Netflix slightly mispredicts what movie you will like. The problem is when this bias moves from the benign to the morally repugnant or manifestly unfair. It is therefore crucial that the design of an algorithm should never be divorced from its intended use.
- When building machine learning algorithms and using their results, we need to tread with caution and think about the context of application.
- Do not let a fancy algorithm...