[CS] Kelly Wagman Candidacy Exam/Nov.5th

[via cs]

This is an announcement of Kelly Wagman's Candidacy Exam.
===============================================
Candidate: Kelly Wagman

Date: Tuesday, November 5th, 2024

Time: 9:30AM- 11:30AM 

Location: JCL 298

Remote Location:https://urldefense.com/v3/__https://uchicago.zoom.us/j/99062510815?pwd=XsoxWcT1Ng14oFmmu1YlEc99ZaZ1oM.1__;!!BpyFHLRN4TMTrA!8SIUWQvUyKauSp2uvXdOq7ek3d1Tq546u4ePYmnnXwftDPPZFlbjIKIHp1PIgfT09DkrCiSMDRZJgK0GgmKHPkay$

Title: A Human-Centered Approach to AI for Science and Healthcare

Abstract: While machine learning has been used as a tool in scientific research for many years, as AI becomes more integrated with sociotechnical systems in science and healthcare a number of questions arise about how people should best interact with these systems. For example, conversational and generative AI hold promise as interfaces but it is still unclear how to best leverage these technologies to improve health outcomes and accelerate scientific research. In addition, some healthcare and science AI models require collecting data from people and there are questions around how to do this effectively and ethically. Little prior work uses a human-centered approach to understand these issues. In this dissertation, I investigate three forms of human-AI interaction in this context. First, I collaborate with a medical clinical trial team to understand to what extent voice assistants and conversational AI can help older adults exercise at home. Second, I study the rollout of an internal generative AI assistant to Science and Operations employees at Argonne National Lab in order to understand both use cases and risks of generative AI in a science organization. Third, I work with climate scientists at Argonne National Lab and a Chicago community non-profit to develop a participatory AI process for involving community members in the data collection for, and use of, hyper-local climate models. Overall, these projects speak to the need for human-centered design in the development of AI sociotechnical systems in science and healthcare from data collection and modeling through end use, risks, and ethical concerns.

Advisors: Marshini Chetty

Committee: Marshini Chetty, Ian Foster, Alex Kale, Mina Lee, Nancy Baym