So the actual numbers are: 5/10 drivers on L2 cars with no L2 experience missed seeing the bear all three times, vs 2/10 drivers with no L2 experience driving a

So the actual numbers are: 5/10 drivers on L2 cars with no L2 experience missed seeing the bear all three times, vs 2/10 drivers with no L2 experience driving a non-automated car. That’s hardly statistically significant. Furthermore, 6/10 drivers with L2 experience saw the bear all three times, as compared to 2/10 drivers using a non-automated car. So a stronger finding (again, ignoring the laughably small sample size) is that experienced L2 drivers driving automated cars are better than inexperienced L2 drivers driving normal cars. In other words, we should put everyone behind a L2 car but require additional training. This is what happens when you design experiments with 3 buckets, 3 tests, and tiny sample sizes. You can draw any conclusion you want.