Abstract
My capstone research addresses a central sociotechnical problem of the AI era: how to balance technological innovation with user privacy in order to sustain public trust. Artificial intelligence systems increasingly mediate everyday life, yet they rely on the collection and processing of personal data. When privacy protections are weak, unclear, or misaligned with user expectations, trust erodes even when the technology performs effectively. To explore this challenge, my team and I developed HooFoundIt, an AI-assisted lost-and-found application designed for a university campus. The platform allows users to upload images and descriptions of lost or found items, using image recognition and keyword matching to generate potential matches, supported by a cloud-based backend that stores submissions and metadata. Because the system handles sensitive information such as photographs, contact details, and timestamps, it raises important privacy concerns. To address these risks, HooFoundIt integrates privacy-by-design principles directly into its architecture, including data minimization, encryption, access controls, and limited data retention. Through this technical implementation, the project demonstrates how privacy can be proactively embedded into AI systems rather than treated as a secondary consideration.
Considering the human and social dimensions of this technology is essential because trust is not determined solely by technical performance. Users must feel confident that their information is handled responsibly and in accordance with their expectations, and even legally compliant systems can undermine trust if they violate contextual norms of information sharing. This research draws on several STS frameworks to analyze that relationship. Privacy-by-design emphasizes embedding ethical protections into system architecture, while Helen Nissenbaum’s theory of contextual integrity defines privacy as appropriate information flow within specific social contexts. Sociotechnical systems theory highlights how technologies operate within broader networks of users, institutions, and governance structures, and user-centered design underscores the importance of aligning development with user expectations. Using a mixed-methods approach, including technical architecture analysis, comparative case studies of AI privacy controversies, qualitative user interviews, and policy analysis of frameworks such as GDPR and U.S. privacy laws, this research finds that privacy-centered design strengthens user trust when protections are transparent, visible, and aligned with contextual expectations. Trust emerges not only from technical safeguards but also from clear communication and institutional accountability. Together, the capstone and STS research show that ethical innovation requires integrating system design, governance, and user perception. HooFoundIt serves both as a functional tool and as a case study demonstrating that privacy is not a constraint on innovation but a foundational condition for trust and long-term technological legitimacy.
