First week policy talk with students, how honest do you get about AI detection limits?
every september i do a frank conversation with my grade 12s about academic integrity before we touch a single assignment. this year i’m rethinking how much i tell them about the limits of detection tools. for fifteen years the message was simple, dont plagiarize, we have turnitin, we will catch you. that message doesnt really hold up anymore given how unreliable ai detection has become. so do i tell them straight up that detection is shaky and lean entirely on trust and process, or keep some ambiguity so the deterrent effect stays intact. i lean toward honesty because kids figure out the gaps within a week anyway and then the whole policy looks like a bluff. but i know some colleagues who deliberately keep things vague. curious how other department heads are framing this conversation this year.
5 Replies
Join the discussion.
Log In to Reply
ive been doing this since long before ai was the topic, cheating scares and plagiarism scares never worked as well as people hoped even with plain old turnitin. give it time, the honesty approach settles in better over a semester than a big scary speech in week one. kids respond to consistency more than fear.
the bluff thing is real. i watched a colleague tell a class detection was basically foolproof and by october three separate students had already found and shared workarounds. now that teacher has zero credibility on the topic for the rest of the year. honesty first, always.
@Tanya Whitfield that lines up with what i was leaning toward. going with full honesty this year, framing it as here is what the tools can and cant do, so lets build assignments where it doesnt matter as much either way.
theres also an equity piece here worth naming out loud with students. the kids most likely to get falsely flagged tend to be esl students and kids who learned formal writing from templates. if you're honest about the limits you're also protecting the kids most at risk of being wrongly accused.
tell them the truth. a policy built on bluffing falls apart the second one kid tests it and tells the whole class. redesign the assessment so detection barely matters. that part isnt up for debate in my department anymore.