Our group decided to analyze the inherent biases in algorithms to determine loan eligibility.
How the Algorithm Makes its Decisions:
In terms of loans, an algorithm would determine whether or not an individual is eligible for one by looking into relevant factors such as the individual’s credit history and income. For instance, it would add up the individual’s total monthly income and analyze it based on how much of it they end up saving and spending. However, the algorithm also tends to take into account other determinants such as the individual’s occupation, economic class, and employee reputation. When it takes particular factors such as these into consideration, it generally serves as the point at which bias begins to creep into such algorithms.
3 Types of Bias the Algorithm may be Susceptible To:
When looking at loan eligibility, the algorithm would be biased against people that have a criminal record, as they are likely to repeat crimes that they have done in the past. The algorithm would also be biased against small companies vs. larger companies because of the income difference and the reputation of the company. There would also be bias against people who have lower incomes, who ironically need loans more, and people that live in poor neighborhoods vs. people that live in rich neighborhoods.
The Effects of Unchecked Biases in this System:
These algorithms would not grant loans to individuals who are the poorest with the lowest socioeconomic status. Those who need loans the most would be singled out and these algorithms have the potential to create a cycle of depth and poor credit because individuals with an existing poor history or low reputation would not be able to get a new loan. This can lead to no financing for purchases such as a new car or home and individuals from poor neighborhoods could continually be stuck where they currently are. Past criminals who legitimately try to get loans could be steered to possibly commit illegal activity if they can’t get a loan because they are stuck. A biased algorithm would overall be very negative towards people who in the past committed a crime or handled credit poorly and it would become like a pithole that would be hard to get out of.