ABSTRACT
The pervasiveness of AI technology is so ingrained within the fabric of our daily experiences that the subtleness of its impact is hidden from the unsuspecting and untrained eye. One of the critical choices in building an algorithm is deciding what outcome it is designed to predict. Bias in AI occurs when an algorithm either has no guardrails or ventures outside of the boundaries of the defined guardrails. When this happens, the algorithm produces results that are systemically prejudiced due to erroneous assumptions. There is no general law in the United States that specifically prohibits algorithmic bias. However, various state, federal, and local laws generally prohibit discrimination, whether by algorithms or by humans, in some contexts, such as credit, education, housing, and employment. Two key legal concepts that animate anti-discrimination laws are: disparate treatment and disparate impact.
