You can’t train people to never make a mistake, particularly when doing highly repetitive work like this. You must build your systems to account for that regardless.
But they do learn and improve.
The models don't (yet).
It might be that the models have been improving in this sense faster than a human child improves.