Weeky seminar: Daniel Brown
October 16, 2024
When: October 17th @ 11am
Where: TMCB 1170
Talk Title: Towards Interactive, Robust, and Aligned AI Systems
We are in an age of “AI everywhere, all at once.” As AI systems become more prevalent in daily life it is increasingly important that their behavior is aligned with human intent and that these AI systems do what we actually want them to do, despite the fact that human intent is often nuanced and hard to formally specify. In this talk I will discuss recent progress towards using human input to enable interactive, robust, and aligned AI systems with a focus on three main topics: (1) how to enable AI systems to estimate human intent, (2) how to make AI systems that are calibrated and robust to uncertainty over human intent, and (3) how robots and other AI systems can efficiently query for additional human input to actively reduce uncertainty and improve their performance.
Dr. Daniel Brown is an assistant professor in the Kahlert School of Computing and the Robotics Center at the University of Utah. He received an NIH Trailblazer award in 2024, was named a Robotics Science and Systems Pioneer in 2021. Daniel’s research focuses on human-AI alignment, human-robot interaction, and robot learning. His goal is to develop AI systems that can safely and efficiently interact with, learn from, teach, and empower human users. His research spans reward and preference learning, human-in-the-loop machine learning, and AI safety, with prior applications in assistive and medical robotics, personal AI assistants, and autonomous driving. He completed his postdoc at UC Berkeley in 2023 and he received his Ph.D. in Computer Science from UT Austin in 2020. Daniel obtained his MS in Computer Science and BS in Mathematics at BYU.