There's probably some precedent with calculators and computers.
Teachers have a really hard job in this case because most writing assignments attempt to exercise thinking, but if AI can write then students will avoid exercising their thinking.
If I were a teacher, I'd probably stop grading long form writing assignments and instead give in-person tests with essay prompts. The tests would have the student draw on work/thinking that should've been done in writing assignments - easy if they did the writing assignments themselves and impossible if they didn't.
I used to give coding exams will full internet access. In general, the students who used AI were easy to tell and they usually got the wrong answers because the AI couldn't take the task and accomplish it from start to finish with no errors.
Last semester, I had 9 students submit almost the exact same script, using methods I had not taught but accomplished the task accurately. When I confronted them, they all denied copying each other, but some admitted to using ChatGPT and one even showed me the prompts he used to get the answer.
I realized that either ChatGPT got better at converting tasks to code, or students got better at prompting the AI.
Either way, I stopped giving coding exams on computer and went back to paper tests where they'd have to write their code out by hand (for fairly simple tasks).
reply
Some professors require students to install special software in their computers to prevent cheating during computer based tests.
On principle, I'm against forcing anyone to install anything on their devices. Especially anti-cheating software that borders the line of malware. Even for my coding assignments, I show them free cloud based options.
reply