Abstract
AI and automation have become more involved with access to opportunities both in employment and education. In my capstone project, I consider how an adaptive scheduling system might improve the scheduling process in universities, while in my STS research I study the problem of bias and discrimination in automated hiring systems. Though both topics might seem different, they both deal with algorithmic systems that shape people’s life perspectives. Both involve decisions based on data that impacts an individual's lives, either by changing their educational career or their job prospects.
Course planning in university is complicated, with students having to deal with prerequisite chains, scheduling problems, and degree requirements with little to no guidance from advisors. Current course registration systems only react to immediate issues, and do not help students proactively plan for their graduation. To solve this problem, I designed a dynamic schedule builder in my capstone project, which builds a personalized degree plan using a greedy algorithm. It puts classes that are required by the degree first while considering prerequisites and the student’s own preferences. Since an algorithm like this affects students’ academic paths, I found that optimizing only for efficiency limited flexibility and student choice. So, I prioritized student preferences and adaptable pathways. This concern about algorithmic systems shaping opportunity also relates to my STS research on bias and fairness in automated hiring.
My STS research focuses on identifying how bias is built into automation tools used during recruitment and how to overcome that bias to ensure accountability and fairness. Automated resume screening software that ranks candidate applications uses historical data to operate, which often reinforces the bias that already exists in society and results in discriminatory practices. In order to study this topic, I use the Actor-Network Theory framework and Social Shaping of Technology concept. These ideas allow understanding that the bias is not the result of an error made by the system but is actually from a larger network of algorithm developers, employers, training dataset, and social conventions. A technical solution cannot solve this problem alone. Instead shared accountability and stronger regulation is needed.
When considered together, my capstone and STS research show the importance of designing algorithmic systems that are efficient and equitable. The adaptive scheduler demonstrates how algorithms can improve decision making in education while my analysis of hiring systems reveals the risks of bias when social factors are ignored. My work shows that technical innovation must be paired with ethical awareness to ensure that algorithmic systems expand rather than limit access to opportunity.
