AI Researcher Dietterich Questions Robotaxi Safety Culture: Too Slow to Fix Problem Behaviors

tdietterich · x · 2026-09-27

Notable AI researcher Thomas Dietterich says he'd love robot cars to succeed (he expects to need one at 80), but argues for broader whole-system analysis and doubts companies have built sufficient safety culture — they seem slow to fix problem behaviors.

In the quoted tweet he adds a sharper point: lower accident rates may partly reflect human drivers being more careful around robot cars, effectively "outsourcing" safety to humans rather than the machines being intrinsically safer.

Original post →

More from Embodied

Embodied channel →