Ujwal Gadiraju | The How, What, and Why of Effective Human-AI Decision Making or How I Discovered this Sisyphean Task

16 MARCH 2023 | 12:30-13:30

The dazzling promises of AI systems to augment humans in various tasks hinge on whether humans can appropriately rely on them. Recent research has shown that appropriate reliance is the key to achieving complementary team performance in AI-assisted decision making. Accurately estimating the trustworthiness of AI advice at the instance level is quite challenging, especially in the absence of performance feedback about the AI system. Moreover, the performance disparity of underlying machine learning models on out-of-distribution data makes the dataset-specific performance feedback unreliable in human-AI collaboration. Through a series of empirical studies operationalizing real-world decision making contexts, we systematically explored the role of explanations, cognitive biases, critical mindsets, and users' affinity to technology in shaping their reliance on AI systems and the overall task outcomes. In this talk, I will share novel insights into why *appropriate reliance* is a non-trivial objective.

Ujwal Gadiraju is an Assistant Professor at the Web Information Systems group of the Faculty of Electrical Engineering, Mathematics and Computer Science (EEMCS/EWI), Delft University of Technology. He is a Director of the Delft AI “Design@Scale” Lab, and a member of the program management team of the TU Delft AI Labs. In addition, Ujwal co-leads the Kappa research line on Crowd Computing and Human-Centered AI at the WIS group. He is a Distinguished Speaker of the ACM, and a board member of CHI Netherlands. Ujwal is currently serving as the Co-Editor for two Frontiers in AI journals: (1) Human-Centered AI and Crowd Computing and (2) User Modeling and Recommendations. He is also an Associate Editor of the Taylor and Francis Behavior and Information Technology journal and the Co-Editor-in-Chief of the Human Computation Journal.