The third contributor to the AI Chasm is what we call the robustness gap. Time and again we have seen that once people in AI find a solution that works some of the time, they assume that with a little more work (and a little more data) it will work all of the time. And that’s just not necessarily so.

This predictive approach turns the problem of driving into something almost perfectly analogous to the ImageNet competition of labeling images. Instead of being shown a picture and needing to categorize it as a dog, cat, flower, or the like, the system is shown a picture from the front dashboard and “categorizes” it as “accelerate,” “brake,” “turn