Agree with Maarten. Don't think of the marks as corrections. Think of them as pointing out differences to the most likely answer. It takes much more computing power (think ChatGPT) to provide a correction tailored to your answer. That's not what this is meant to be.

With the differences pointed out to you, and other possible variants listed, it is up to you to determine how you grade your reply. That's why these are self-graded exercises.

I agree. Marking your own exam is as strange as participation trophies!

I would say ‘careful what you wish for’ ! 

With all the minor things you could ‘get wrong’, do you really want a machine to determine whether you know the material ? An upgraded machine with even greater artificial ‘ non-intelligence ‘ ?

Even on simple one or two word fill in the blanks, there are more than enough issues with bot correction.