Can Machine Learning Identify Criminals Just by Looking at Their Faces?

In 2016, AI researchers Xiaolin Wu and Xi Zhang posted a paper to a non-peer reviewed depository that explained their plan to use machine learning to recognize criminals just by looking at their faces. They claimed to discover that “some discriminating structural features for predicting criminality have been found by machine learning.” There was an immediate backlash. According to Wu and Zhang, critics called the project racist. The researchers defended their stance, admitting that some of the language they used did have negative connotations due to oversights in translating their research into English, but overall standing by the project. This case study examines the ethics of their research. The analysis provided by the author argues that the kind of research under discussion in this case study is not ethical to carry out.