Systematic Bias

This academic study explores how prejudice becomes embedded in automated systems. It approaches the subject from three directions: bias as it appears in cultural works, the magical and mathematical metaphors that circulate within technology culture, and the use of psychometric measurement in workplaces.

What a model produces depends on the data it receives, but also on the connections and presumptions built into its design. Many of those hidden premises arise from social and cultural factors rather than from physics or mathematics. The book asks how human prejudices find their way into data models and surveys the strategies that have been proposed to counter bias in algorithmic results.

It is written for scholars and students in many fields, as well as policymakers, journalists and general readers. Drawing on research from communication studies, art and new media, it offers a wide-ranging look at algorithmic bias.