Date: October 31, 2024

Time: 9:30am-12:30pm EST 

Location (Zoom only): Zoom Link | Meeting ID: 894 4124 6737 Passcode: 325680

Upol Ehsan 

PhD Candidate in Computer Science

School of Interactive Computing

Georgia Institute of Technology 

Committee: 

Dr. Mark O. Riedl (advisor) – School of Interactive Computing, Georgia Institute of Technology 

Dr. Munmun De Choudhury – School of Interactive Computing, Georgia Institute of Technology 

Dr. Sashank Varma – School of Interactive Computing and School of Psychology, Georgia Institute of Technology 

Dr. Q. Vera Liao – Microsoft Research 

Dr. Michael Muller – IBM Research 

Summary: 

If AI systems are going to inform consequential decisions such as deciding whether you should get a loan or receive an organ transplant, they must be explainable to everyone, not just software engineers. Despite commendable technical progress in “opening” the black-box of AI, the prevailing algorithm-centered Explainable AI (XAI) view overlooks a vital insight: who opens the black-box matters just as much as opening it. As a result of this blind spot, many popular XAI interventions have been ineffective and even harmful in real-world settings.  

To address the blind spot, my dissertation pioneers and operationalizes the domain of Human-centered XAI (HCXAI), a sociotechnically-informed human-centered sub-field of Explainable AI. 

Thesis statement: With a focus on non-AI experts, this dissertation demonstrates how Human-centered XAI: 

1. expands the design space of XAI by broadening the domain of non-algorithmic factors that augment AI explainability
2. enriches our knowledge of the importance of “who” the humans are in XAI design  
3. enables resourceful ways to do Responsible AI by providing proactive mitigation strategies through participatory methods 

It contributes 1) conceptually: new concepts such as such as Social Transparency that showcase how to encode socio-organizational context to augment explainability without changing the internal model; 2) methodologically: human-centered evaluation of XAI, actionable frameworks, and participatory methods to co-design XAI systems; 3) technically: computational techniques and design artifacts; 4) empirically: findings such as how one’s AI background impacts one’s interpretation of AI explanations, user perceptions of real AI users, and how AI explanations can negatively impact users despite our best intentions.  

Impact: Epistemologically, the dissertation expands the XAI discourse from an algorithm-centered perspective to a human-centered one. In terms of real-world impact, the work has informed RAI policies at influential institutes like the United Nations and Mozilla Foundation. Seven Fortune100 companies have adopted the technology from my work, improving trust calibration and AI explainability of 3 million users. The work serves the mission of making AI systems explainable and responsible so that people who are not at the table do not end up on the menu.