Abstract
During the summer of 2022, the Finance Platform team in an American banking firm headquartered in Virginia noticed that its data preview and download feature was malfunctioning. More specifically, miscommunication between the feature’s backend microservice and the frontend User Interface (UI) component caused erroneous behavior in the application when users attempted to preview larger and more complex datasets. Because this feature’s user base consisted primarily of the firm’s employees due to it comprising a portion of their internal application, my internship focused on replacing the existing feature with an improved version. My technical report details the steps that my internship team and I took to create an upgraded data preview and download backend service and its respective UI component. We hoped that doing so would rectify the previous feature’s shortcomings.
My summer internship, which also served as my technical project, was managed using agile methodologies. Moreover, its technology stack included the Spring framework and Apache Spark for the backend, the Angular framework for the frontend, and Amazon Web Services (AWS) to satisfy data storage requirements. Utilizing the provided combination of tools and technologies allowed us to efficiently replace the existing service with one that incorporated a more comprehensive set of endpoints (e.g., preview, download, download zip) and a redesigned preview response object. To maintain consistency with the company’s coding practices, our team also created a UI component using Angular to facilitate a more user-friendly experience when previewing and downloading the selected datasets.
Although the newly created service and UI component performed slower than we had originally anticipated, they allowed users to successfully preview and download a more diverse range of datasets with minimal challenges. However, potential next steps would include extensive regression and performance testing, improving performance time by leveraging newer technologies, and integrating this feature into the firm’s production environment.
Despite my technical project focusing on modifying the existing data preview and download feature in the firm’s internal web application, this company also offers a myriad of other technological products and services to its customers. Even though I was only able to interact with a subset of the firm’s products, I was the most intrigued by Eno, its chatbot. Over the past few decades, advancements in fields including Artificial Intelligence (AI) and Machine Learning (ML) have helped revolutionize how businesses communicate with their customer base. More specifically, a majority of these companies across various industries have been rapidly transitioning away from customer support teams and into adopting chatbots. While chatbots have undoubtedly aided businesses by increasing productivity and minimizing cost, the STS paper investigates how generative chatbots in the banking industry impact the retention rate of low-income customers.
Chatbots are merely computer programs that are designed with the intent of simulating and comprehending ordinary human conversation. Although earlier chatbots such as ELIZA, PARRY, and ALICE typically exhibited limited functionality because of their underlying rule-based technology, the progression of computer science has allowed the creation of generative chatbots such as Capital One’s Eno and Amazon’s Alexa among many others. In an attempt to narrow down the scope of the research topic, the STS paper solely focuses on generative chatbots.
While chatbots have been widely popularized and even idolized in certain circumstances, this paper draws on various research (e.g., academic papers, research articles) to examine how aspects of a chatbot including functionality, credibility, and emotional objectivity are perceived by different social groups. Due to the nature of my topic, the social groups are stratified by both income and race. Moreover, a majority of the research focuses on the relationship between chatbots and the low-income community. To better understand the relationships between chatbots and the impacted social groups, the STS paper relies on the Social Construction of Technology (SCOT) framework. Given that this framework revolves around the concept that society shapes technology, it effectively supplements my analysis of how this technology is being continuously adapted in response to both the approvals and criticisms of different social groups such as the low-income population.
Finally, a paper that concentrates on chatbots and their implications within communities would be incomplete without discussing the new chatbot that has taken the AI and ML fields by storm – ChatGPT. To acknowledge chatbots of this caliber, my research paper briefly examines how ChatGPT can potentially modify the banking landscape. Additionally, it touches on what this may mean when considering the ever-evolving relationship between the low-income population and this technology.
