What will make autonomous cars better is that they can take all of the input from thousands of blowouts, add "training" from actual experts, and load the resulting techniques into every car on the road, ready for execution the instant something happens.
There are two things where human beings far outpace AI in competence. The first is the ability to recognize when inputs just simply don't make sense and thus the intended algorithm is going to return garbage results. A related, but important issue (particularly if you talk about networking cars together) is being able to recognize malicious inputs.
The second issue is that humans are good at higher-order planning. An example as it relates to driving is what we do when we can't tell the state of the traffic light (direct sunlight, say). Humans naturally look for ways to indirectly infer the state: the status of the traffic light in the cross direction, the current disposition of cars around the intersection.
Autonomous cars will only be better at handling these situations if the engineers are aware that these situations need to be handled and take measures to make the cars aware of them. And history suggests that is sufficiently unlikely to happen to make me fearful of the future.
And yes, we can deduct light color from other clues, but we can also go through a red light because we were preoccupied with our cellphones. AV will on the other hand have access to multiple cameras, looking at the light from multiple angles, possibly with some polarizing filters that will make this non-issue.
Which rapidly reduces the MTBF at which point you're stuck with an overpriced manual drive vehicle.
Autonomous cars will be much better at handling these types of situations - in your example, the machine will know what's going on few milliseconds after it happens and will be able to adjust speed and direction almost instantaneously. All you need is lots of sensors.