Weapons of Math Destruction (2016) by Cathy O’Neil looks at how sets of modelling equations can discriminate and cause other problems.
O’Neil looks at how financial models that were not correct helped cause the 2008 Financial crisis, how recidivism models effectively discriminate against US minorities, software that automatically writes up people who are late, how allowing advertising to target poor people is unfair, how policing algorithms may be unfair, how IMPACT teaching scoring is unfair, how car insurance based on models can be discriminatory and unfair, how recommendation and search algorithms have considerable power.
The book brings up some interesting questions but also O’Neil has a fairly strong point of view that comes through strongly. She also clearly downplays or ignores the upside to many of the algorithms she writes about and also sometimes doesn’t consider the alternatives. She also fails to contemplate what competition in the market can do.
The book definitely has some interesting examples of what algorithms and machine learning can do and provides some interesting material to think about. But the bias and lack of balance lets the book down.