As AI agents become more capable, their role in the data science workflow is evolving—from mere accelerators to genuine collaborators. In this talk, Dr. Jason Hillary explores the realities of human-agent collaboration in AI and data science teams: where it shines, where it stumbles, and where it’s heading. Drawing from real-world implementations and research lab trials, the session highlights success stories—such as code generation, model debugging, and dynamic data exploration—as well as persistent challenges like agent misalignment, opaque reasoning, and brittle context management. This talk is for anyone interested in building or working alongside AI agents not just as tools, but as teammates.
Technical level: Technical practitioner
Session Length: 40 minutes
Supported by
