Racially Biased Algorithms

Abstract
The role of algorithms in big data is to help with the analysis of that data for the purposes of making intelligent decisions by the big technology companies such as Facebook, Google, Amazon, Twitter, and others. The need for the algorithms in big data is quite simple, the data is too big to manage within reasonable time frames for intelligent analysis. Therefore, algorithms help to simplify the process. The definition of an algorithm in layman’s terms is just a way for a computer to follow a certain set of rules to accomplish a specific goal. The rules are normally variables, and the goals may also be variable based on per-determined factors. Usually, a person or team decides about the factors and variables. In theory, big data algorithms should be something that is tweaked on a regular basis to improve desired outcomes.

Modern algorithms use both Artificial Intelligence (AI) and Machine Learning (ML) to analyze data. Machine Learning (ML) which is a subset of AI uses mathematical models which are designed to learn and adapt, which can be done in a supervised or un-supervised manner to make decisions that are in alignment with the goals of the organization. Since algorithms are absolutely necessary in analyzing big data, it is important that the models, the training of the models, and data used all work together to generate objective results.

Problem Statement
Discrimination has been an issue in our American society since the founding of our county. There are many laws in place to battle against discrimination such as the Civil Rights Act and voting laws as well as the American Civil Liberties Union (ACLU) and the National Association for the Advancement of Colored People (NAACP). However, discrimination still prevails in many ways. This can be seen is red lining, mortgage discrimination, credit inequality, job inequality, education, and the criminal justice system. As technology improves on a yearly basis, the problems with discrimination have become more sophisticated. This means that for those who want to discriminate can do so more easily with technology and for implicit bias which is baked in or coded into computer algorithms make to more difficult to track and find its source, but still have the same net result of discrimination. So, with technology discrimination can be done directly or explicitly, or it can be gone implicitly; the result is still discriminatory, just with more sophistication based on its technology.

This leads us to very specific issues such as with online searches and specifically racial discrimination in online ad delivery. We can also see this issue in the criminal justice system in algorithms deals with racially biased recidivism. Then, we have issues in housing and banking where people can be discriminated against without even know it. Another huge problem is facial recognition used by big tech companies and some law enforcement agencies. Facial recognition can be a great tool if used properly, but it has inherent problems which can lean towards misrepresentation and bias. Algorithmic bias in our society is ultimately ubiquitous.

Significance of the Research
This research will bring to light those algorithms in our society that have implicit and explicit bias programmed into them and that the problem is ubiquitous. This is important because as a society we as citizens have a need and a right to have the algorithm codes that affect our daily lives be publicly available and made to be legally fair and free of discrimination.

Research Questions
RQ1 – Do artificial intelligence/machine learning algorithms lead to implicit or explicit discrimination in general?

RQ2 – Does technology discriminate against minorities in the areas of education, housing and employment using online advertising?

RQ3 – Is implicit bias encoded into the artificial intelligence/machine learning algorithms that Google Adwords uses which has the net result of discriminating against non-white United States citizens in online ad delivery in the areas of education, housing, and employment?