BIAS NOTIFICATION DUTY.

Algorithms are often tainted with bias that could negatively affect subjects' lives in many aspects. But while policymakers and scholars tend to focus their attention on methods to Jix the bias within the algorithm or its output, they neglect a crucial piece of the puzzle: when companies discover bias and fix it in the algorithmic output, they do so behind the scenes. While such a move might debias the output, society remains in the dark about its existence or how it might have affected the outcome. Sociely thus misses a unique opportunity to study social and cognitive biases that shape our civil rights and liberties today-and might become more important in the future with the rise of automation. We thus propose a new mechanism to aid in the study of bias and its elimination, not only in algorithms but also in society. We call it Bias Notification Duty (BND). This article lays the groundwork for this new social-legal responsibility by offering a dilfevent viewpoint on bias, algorithms, and society. We propose that the state imposes a duty (BND) on companies and their employees to, when bias is found, provide a notice to a selected governing body that will further study it and, upon evaluation of its source and impact, will notify those who are both directly and indirectly affected by it accordingly. Affected parties may include other companies that could be subjected to similar bias, individuals that were potentially harmed by it, researchers in relevant fields, and society as a whole. Upon setting the proposal, we examine the rationales behind it and discuss its hurdles. With such acknowledgment, we argue that BNDs are an important legal tool that should be used to aid in de-biasing our society, especially if the world will eventually be algorithmically driven and bias will be even more obscured by technological companies. [ABSTRACT FROM AUTHOR]