Designing a Requirements Engineering Web-Based Platform; A Social Construction of Technology Analysis of the Apple Credit Card AI Bias

In both my technical project and my STS research paper, I examine the power and complexity of artificial intelligence systems, with a particular focus on how human decisions and institutional priorities influence technological outcomes. Though the projects differ in application, they share a central theme: technology is not neutral. My technical project leverages NLP and machine learning to improve how engineering teams manage and validate system requirements. My STS paper, meanwhile, applies the Social Construction of Technology (SCOT) framework to reveal how social groups shape the development and deployment of biased AI systems. These projects demonstrate how technical solutions must be developed with critical awareness of their broader social implications. In my technical project I constructed a platform to streamline requirements engineering; so, to gain insight into how more robust requirements from a variety of social contexts are critical to technological success, my STS research examines the social construction that led to the failure of the Apple credit card scandal.

The technical report presents a web-based platform designed to automate and enhance the requirements engineering process using Natural Language Processing (NLP), Large Language Models (LLMs), and Latent Dirichlet Allocation (LDA). The platform enables users to extract structured requirements from documents, validate compliance with INCOSE standards, and visualize requirement groupings. This tool aims to reduce errors, improve efficiency, and lower costs by addressing long-standing challenges in how engineering teams define system needs. The system reflects current best practices in full-stack development and applies AI tools to a core stage of the software development lifecycle. The project concludes that by embedding machine learning into technical documentation workflows, we can increase both speed and accuracy in software project planning.

My STS research analyzes the Apple Credit Card algorithm using the SCOT framework to argue that algorithmic bias is not simply a technical flaw but a socially constructed phenomenon. By examining the priorities of relevant social groups, particularly Goldman Sachs and regulatory agencies, the paper shows how historical discrimination, opaque algorithm design, and reactive oversight contributed to biased outcomes. SCOT’s concept of interpretive flexibility illuminates how different stakeholders shaped the AI system’s trajectory, ultimately resulting in gender-discriminatory credit scoring. The analysis critiques the assumption that technical refinement alone can resolve bias, emphasizing instead the need for structural reforms in AI governance and increased institutional accountability.

Working on both projects deepened my understanding of the importance of designing with ethics in mind. My technical work taught me how easy it is to overlook sources of bias and make errors when focusing solely on performance metrics or engineering efficiency. Additionally, being able to keep track of various technical and ethical requirements is paramount to creating less biased technologies. At the same time, my STS research reinforced that even well-intentioned AI tools can perpetuate harm if social context is ignored. In future projects, I plan to apply the SCOT perspective to anticipate how design decisions might be interpreted differently by users and regulators, and to ensure that systems remain transparent, auditable, and aligned with public values.

School of Engineering and Applied Science
Bachelor of Science in Computer Science
Technical Advisor: Rosanne Vrugtman
STS Advisor: Benjamin Laugelli
Technical Team Members: Maria Salamone