Is Outlook going to come with an AI that helps an engineering manager succinctly express the pros and cons of two competing approaches to a nontechnical audience and then justifies the decision to go with one over the other? What if these approaches are novel or otherwise something an expert system can't "know about" (or look up on the Wiki)?
Regardless, IMO this is where the discussion becomes too hypothetical:
• Right now, AIs are bad at writing.
• If a student is able to masterfully craft a prompt which causes today's bad AIs to produce good writing, that student is talented, and I suspect they actually did a significant amount of work.
• If today's bad AIs are able to produce assignments which receive passing grades with minimal prompting, something is wrong with the assignment. Teachers should assign work that isn't so rote a computer can do it.
• If, in the future, AIs are able to write strong, well-reasoned essays about novel concepts, students won't need to learn how to write. They may not need to learn much of anything†, because I don't think there will be much space left for humans in the workforce.
-----
† "Need" is a key word here; I do believe in learning for its own sake, and apparently we'll have lots of time on our hands. Cool story btw, learning how to develop film in a dark room is lots of fun, would recommend.
AI assisted technical writing is already a thing. You might need to give it more input and do more cleanup on what it spits out, but it can still greatly reduce the time and effort needed. There's no reason to believe the technology will stop improving anytime soon.
