The people in charge of developing self-driving cars keep backpedaling their optimistic predictions. It's looking like a classic case of AI overreach.

Sure, but they seem possible by 2030, and that's the earlier that I'd expect any change in driver education to have a notable impact on auto deaths/