This talk explores the evolving dynamics of human+AI collaboration, focusing on the concept of the human as a “hazy oracle” rather than an infallible source. It outlines the journey of integrating AI systems more deeply into practical applications through human+AI cooperation, discussing the potential value and challenges. The discussion includes the modeling of interaction errors and the strategic choices between immediate AI inference or seeking additional human input, supported by results from a user study on optimizing these collaborations.

About the Speaker

Jason Corso is a Professor of Robotics, Electrical Engineering, and Computer Science at the University of Michigan, and Co-Founder / Chief Scientist at AI startup Voxel51. His research spans computer vision, robotics, and AI, with over 150 peer-reviewed publications.

Not a Meetup member? Sign up to attend the next event:

https://voxel51.com/computer-vision-ai-meetups/

Recorded on May 8, 2024 at the AI, Machine Learning and Data Science Meetup.