Cathy O’Neil’s book Weapons of Math Destruction explores the dark side of Big Data. In particular, it explores a number of case studies that highlight the damage that can be done through the misuse or outright abuse of statistical models, with an emphasis on the socioeconomic impact.
Sometimes, it comes down to bad statistics. Other times, the models are opaque and not accountable. In other words, the models are not interpretable and their predictions are not being properly validated. In some cases, externalities are at play, where the interests of the people designing the models are different from those of the people the model affects (ex: predicting prison recidivism).
Usually, people on the lower end of the socioeconomic spectrum are the victims of flawed statistical models. For instance, models used to screen job or loan applicants, or to allocate police officers to neighborhoods based on crime likelihood.
I strongly recommend this book for any data scientist. It presents a discussion on ethics in Big Data and analytics that isn’t present enough in a lot of fields. Privacy aside, Big Data is a dangerous tool, and we need to be aware of the damage it can cause.