OhSnap!DragonMar 24
Just got back from Spring Break, and my to-do list this week includes failing three students in my freshman-level academic writing class for using AI to write their papers. I used white font to implant some "invisible" instructions on the assignment sheet, so when they upload it or cut-n-paste it into an LLM, the bot follows the secret input, et voila!
BUSTED!
I'll be so glad to retire in May and never have to deal w/ this BULLSHIT ever again.
#teaching #writing
#HigherEducation
#University
Show threadProfessor_StevensMar 27

@DrOinFL

That trick catches the guilty, but it doesn't discourage the crime.

I use a blend of homework and in-class work. Students must score at least 60% on BOTH to pass. The smart ones realize that this means doing the homework themselves will help them pass the in-class work. The not-smart ones get 95% on the homework by using LLMs, and 55% on the in-class work.

You can see why this prevents the crime.

Show threadMichkovMar 29

@Professor_Stevens @DrOinFL

I suppose a warning at the beginning of the course and maybe a "x/y people failed this course because of LLM use" statistic would do the trick.

Show threadProfessor_Stevens

@Michkov @DrOinFL

That would require at least an iota of humility for it to work. Students have, ant most, a scintilla. An iota is too much to ask of them.

Mar 29 at 1:58amWeb