Stephen Downes

According to Philippa Hartman, "AI’s best work is reachable, but takes a lot of user time, energy and and skill." This is the sort of output you get when you as AI to push back, when you don't setting for the quick solution, and when you tell AI to stop and ask when it's uncertain. You'd think these are in every AI user's toolbox, but not according to her study of learning designers. She calls this 'extra' work "the AI steering tax", but I just call it 'working the problem's.