ABSTRACT

An individual from a particular group is deemed to inherit the presumed attributes of the group (a stereotype), and occasionally,the attributes of an individual from an unfamiliar group are deemed to be representative of that group and defines it. This chapter examines the psychology of biases of race, gender, religion, and so on, and takes the view that they are the result of normal human processes of category induction from information in the surrounding culture. This is tapped into in the implicit associations test (IAT). While responses to racism scales may show no negative feelings towards African Americans, the IAT suggests there are unconscious ones. Diversity training has been established in many institutions to combat biases but, since they constitute a heuristic or thinking habit, they are difficult to overcome. To circumvent ‘faulty’ human biases, machine learning systems have been used on the assumption that they are neutral. However, various studies have shown that they produce the same results as humans as they were trained on biased human data. Since human biases are also the result of cultural transmission, the bias is in the data rather than in a faulty cognitive system.