Lost in Compression: Who’s Heard and Who’s Blurred in Digital Voice Communication?; Analyzing the Effectiveness of Using Artificial Intelligence in Customer Service

Both of my projects explore how artificial intelligence and digital systems can unintentionally reinforce biases that impact users, particularly in customer service contexts. My STS research analyzes the effectiveness of AI in customer service through the Social Construction of Technology (SCOT) framework, highlighting how different stakeholders influence and interpret the ethical challenges of AI adoption. It focuses on broader concerns such as data privacy, transparency, algorithmic bias, and the human implications of automation. Meanwhile, my technical project, "Lost in Compression: Who’s Heard and Who’s Blurred in Digital Voice Communication?," investigates a specific technical source of bias: how digital voice codecs can degrade the speech quality of certain demographic groups, particularly non-native English speakers and female voices. By analyzing objective audio quality measures across different compression algorithms (Opus, Codec2, and AMR), my technical work exposes how the design choices behind widely used communication technologies can systematically disadvantage certain users. Together, these projects provide a comprehensive understanding of the hidden biases that can arise both at the algorithmic level and at the user experience level. The technical work identifies measurable disparities caused by digital infrastructure itself, while the STS paper addresses how these technical biases intersect with social perceptions, trust, and broader ethical concerns in AI-driven customer service. This dual approach underscores the importance of integrating ethical considerations into both the design and deployment phases of technology development, ensuring that AI systems and digital communication tools serve diverse populations fairly and equitably.