AI makes it far less tedious to write out that test code [...]

Completely disagree.
In my experience, LLMs constantly generate bad code that needs to be thoroughly checked, to the point that writing by hand is more practical.