I had brought up the consistency issue in the test patch thread before patch Y and I think it was mentioned in other threads (not in this subforum - mostly in General) after it.
Superhuman consistency (and inconsistency) came with Y, previous to it the AIs had their own knowledge files and had the potential to vary between them quite a bit depending on their previous experience at overtaking, etc. A common way of training them was to put the faster AIs at the rear of a field of slower AIs.
Now it's sort of funny to watch 20 AIs in a line take a turn by breaking and locking at exactly the same place and turning the car following exactly the same line - it reminds me of a conga line. But at least they can pit-in now and they're quite quicker than they used to be which makes them marginally more interesting when practicing on a track.
Superhuman consistency (and inconsistency) came with Y, previous to it the AIs had their own knowledge files and had the potential to vary between them quite a bit depending on their previous experience at overtaking, etc. A common way of training them was to put the faster AIs at the rear of a field of slower AIs.
Now it's sort of funny to watch 20 AIs in a line take a turn by breaking and locking at exactly the same place and turning the car following exactly the same line - it reminds me of a conga line. But at least they can pit-in now and they're quite quicker than they used to be which makes them marginally more interesting when practicing on a track.