What is bias in machine learning?

- In the 17th century, Europeans believed all swans were white. They were so certain of this fact that when referring to something impossible, they would say it was like a rare bird in the lands, and very much like a black swan. You see, back then the only swans Europeans had ever seen were white. And from this they came to the simple conclusion that all swans must be white. Their own observations, their data, was biased to the environment they were in, Europe. And in that environment, all swans were indeed white. Then in 1697, a Dutch explorer traveled to Western Australia and discovered a black swan. And with that discovery, people had to reevaluate their understanding of the world. The problem here is one of inference. We humans like to infer rules based on our observations of the world. We observe a bird flying and infer that all birds can fly. We observe white swans and we assume all swans are white. These inferences are all based on our biases. I'm a 40-something white male…