Written by Cathy O’Neil
Machine Learning and other mathematical algorithms are more and move taking over decision making. Whether someone is given a loan or an insurance policy. If someone gets fired because of not achieving the required efficiency based on the algorithm’s requirements or if someone gets the job in the first place. If someone might continue to commit crimes if we keep them on the street. If someone is showered with expensive, for-profit college advertisements or other high-interest, predatory loans. The decisions these models make are considered objectives and not negotiable. However, they are at a severe risk of a nasty feedback loop.
In this book, Cathy describes many cases of this happening in the world right now. She has witnessed first hand how models are being given too much power and how not enough energy or thought is being invested into designing and developing ‘unbiased’ models.
The book does have a clear tendency towards the negative consequences of machine learning models. Little to no attention is given to the positive and innovative use cases that have come from these models. The books also feels a little long by the sheer number of cases she brings up. After the 10th case of biased models the message has gotten through clear enough and adding additional examples felt a little redundant.
I’d say: Read it