[ex-aws-dude]

Right but does this just rely on some human "brainstorming" a bunch of edge cases?

It just strikes me as neverending edge case wack-a-mole

A human doesn't need to see tons of examples of tornados and elephants to know to stop the car

Doesn't that indicate some fundamental difference between the model and a human driver?

[sixo]

It might be "never-ending", but you're going to encounter edge cases in approximate proportion to the rate at which they actually occur. Anyway, the hope would be to learn behaviors which generalize, not to respond to each edge case ad-hoc; the edge cases provide out-of-sample tests of generalizability.

[Mawr]

Neither does the car — it won't drive into what LIDAR sees as a wall. But stopping is not good enough, it needs to be able to navigate the obstacle as well.

Also, even if the car behaved perfectly anyway, these scenarios are useful for testing — validating that the expected behavior happens.