Designing People-Centered AI

Speaker Name: 
Min Kyung Lee
Speaker Title: 
Research Scientist
Speaker Organization: 
Carnegie Mellon University
Start Time: 
Monday, March 11, 2019 - 11:15am
End Time: 
Monday, March 11, 2019 - 12:15pm
Katherine Isbister
Abstract: Artificial intelligence (AI) is transforming work and society at many levels. Automated decision systems are beginning to manage worker routines, tasks, and incentives, and even who is offered jobs or small business loans. AI is being applied in the criminal justice system and government programs. Such transformations promise efficiency and data-driven insights, but they risk unfair decision-making and harm to individual autonomy. To better understand these benefits and risks, my goal is to examine the tradeoffs in algorithm design and the assumptions behind claims of best practice. I combine design, computer science, and social science research to study how algorithms are being applied in real life and how they affect people.
In this talk, I present two studies of people’s responses to ridesharing and task allocation algorithms. Algorithms rely on simplified assumptions and computational guarantees, whereas social contexts involve human emotions, values, norms, and routines. Our studies offer empirical evidence of how the gap between algorithms and social contexts can be manifested, and how it can harm individuals’ trust, perceived fairness, and acceptance of algorithmic work practices. I will then describe my current design efforts to build a participatory framework to mitigate these risks. I aim to support multiple stakeholders in building an algorithmic service collectively so that the resulting algorithm reflects the community’s needs and values. I co-developed and applied this framework to build a fair and efficient algorithm in the context of on-demand donation allocation, working with stakeholders of the nonprofit 412 Food Rescue. Our findings underscore the importance of stakeholder participation in AI design, and contribute new ways of enabling algorithmic fairness, awareness and human-centered design of AI systems. By bridging the gaps between AI and human contexts, we can make algorithms that work fairly and promote wellbeing.
Bio: Min Kyung Lee is a research scientist in Human-Computer Interaction in the Machine Learning Department and the Center for Machine Learning and Health at Carnegie Mellon University. Dr. Lee has conducted some of the first studies that empirically examine the social implications of algorithms' emerging roles in management and governance in society, looking at how people perceive algorithms and how we can design fairer and more trustworthy algorithmic services that work in the real world. Her current research is inspired by and complements her previous work on social robots for long-term interaction, seamless human-robot handovers, and telepresence robots. Dr. Lee is a Siebel Scholar and has received the Allen Newell Award for Research Excellence, research grants from NSF and Uptake, and five best paper awards or honorable mentions in venues such as CHI, CSCW, DIS and HRI. She is an associate editor of the ACM Transactions on Human-Robot Interaction, and is co-organizing Special Issue on Unifying HCI and AI for Human-Computer Interaction Journal. Her work has been featured in media outlets such as the New York Times, New Scientist, Washington Post, MIT Technology Review and CBS. She received a PhD in Human-Computer Interaction and an MDes in Interaction Design from Carnegie Mellon.