Chapter Two. What the Machine Actually Learned.
When the Data Is the Problem
Machine learning doesn't learn what you want it to learn. It learns
what the data says, including everything the data says that you wish it
didn't.
In 2014, a small team of engineers at Amazon had what seemed, at the
time, like a very good idea.
Hiring was a problem at Amazon. It was a problem at every tech company,
but at Amazon it had a particular shape. The company was growing at a
rate that made conventional rec