Think IBM training manual had a quote about not being able to hold a computer accountable, so you cant let them make management decisions. Basically management will stick around so somebody can be held to blame if AI slop breaks