In their article titled “Discrimination in the Age of Algorithms,” Joh Kleinberg, Jens Ludwing, Sendhil Mullainathan, and Cass R. Sunstein explore how algorithms can be used to determine whether discrimination occurs during a decision-making process. Often, discrimination occurs within the cognitive sphere, making it difficult to trace and prove. However, with appropriate systems in place, it could become possible and easy to examine and interrogate decision-making processes to identify incidents of discrimination.
Section 1 aims to introduce the concept of algorithms, as can be used to determine whether discrimination has occurred. The introduction is an exploration of the difficulties of determining and proving the occurrence of discrimination. Kleinberg et al. assert that proving discrimination is a challenging task because of human decisions’ opaque nature (2). Humans dissemble, lie, obfuscate, and may not even know when they discriminate against other people. However, proving discrimination is bound to become easier if the algorithm detection process is developed and actualized.
The second section of the paper, “The Law of Discrimination: A Primer,” presents an exploration of the basis of discrimination law. Kleinberg et al. highlight two problems associated with the discrimination law: disparate treatment and disparate impact (6). The law forbids disparate treatment but remains impartial about the disparate impact (Kleinberg et al. 5). Discrimination involving disparate treatment arises in situations where people of a particular gender or race gain favor over others, making it easy to detect. Problems arise in proving disparate impact discrimination because it is manifested in daily life through stereotypes and biases.
The third section of the article, “Anti-Discrimination Law for Humans,” presents an exploration of existing laws meant to curtain discrimination. In essence, decisions in all settings involve the application of an implicit or explicit screening rule. The complexity of these screening rules makes it difficult to establish whether a decision is discriminatory (Kleinberg et al. 10). Nevertheless, discrimination is inscribed in human cognition and biases that society presumes to be the norm. In such cases, it becomes difficult to prove discrimination because the perpetrator is not even aware.
In overview, sections 1, 2, & 3 provide a vivid background to the challenges that arise when it comes to proving discrimination. It emerges that discrimination is cognitive and does not always leave a trace to serve as evidence. However, carefully designed algorithms could be used to identify and prove the discrimination occurred.
Reference
Kleinberg, Jon, et al. “Discrimination in the Age of Algorithms.” Journal of Legal Analysis, vol. 10, 2018, pp. 1-45.