Generative Artificial Intelligence: The Critical Role of Data on AI Outputs; An Analysis of Sociotechnical Networks Surrounding the Boeing 737 MAX Disaster

In my technical project, I analyzed how the quality of training data influences the performance and reliability of generative AI systems. I conducted a meta-study with existing research, finding that poor-quality, biased, or synthetic data can harm AI outputs, causing hallucinations and loss of diversity. In my STS research paper, I examined how the interactions between human and non-human actors in the Boeing disaster led to catastrophic failures when important connections weakened and eventually broke down. Although both papers focus on different industries, they explore how systems fail when critical inputs are faulty. In my technical project, data quality served as the input that determined the performance of the system. In my STS project, the quality of actor interactions acted as inputs into the aviation safety network. In both cases, failures were not the result of a single malfunction, but the gradual weakening of a system through overlooked vulnerabilities. Both projects emphasize the need for attention to system inputs to avoid future issues.

School of Engineering and Applied Science
Bachelor of Science in Computer Science
Technical Advisor: Rosanne Vrugtman
STS Advisor: Kathryn Webb-Destefano, Gerard Fitzgerald
Technical Team Members: Henry Chen