Presenter: Dr. Kristina Kupferschmidt, candidate for a tenure-track position in the School of Mathematical and Computational Sciences 

Teaching Talk: Friday, April 19, 9:30--10:30 am, Cass Science Hall 101

Research Talk: Friday, April 19, 1:30--2:30 pm, Cass Science Hall 101

Title: "Back to the building blocks: Making AI more human-centric through data-centric practices" 

Abstract: The growing popularity of artificial intelligence (AI) and machine learning (ML) presents tremendous opportunities for high-stakes fields such as medicine and public policy. However, the risks associated with their use must be properly quantified and mitigated. Human-centered AI (HCAI) is an approach that applies design thinking, with a heavy focus on collaboration with experts, to facilitate the responsible development of AI systems. HCAI typically considers many pillars including fairness, explainability, and usability; however, the current literature often fails to recognize the complex interactions between these pillars, and recommendations are often theoretical in nature. 

To address these challenges, my ongoing research has focused on developing practical tools and studying real-world examples that demonstrate effective integrations of HCAI across different domains. My future research program will continue to build on this work by leveraging modern ML technologies, such as large language models (LLMs) and multimodal transformers, to provide additional context to models designed for high-stakes scenarios. These settings, which often rely on expert insight to consider complex contextual factors such as diagnostic risks or ethical values, highlight the limitations of current AI systems. By enhancing the ability of ML systems to process and interpret the contexts in which they operate, my research aims to significantly improve the transition of novel ML tools from the experimental stage to practical, reliable applications. The potential benefits expand beyond just improving technical performance but also include building user trust in AI systems.

Teaching and research talks are open to the UPEI campus community.