The AI Signal Check

A HARVEST KERNEL WORKING SESSION

The AI Signal Check

Take one course and read it the way a student does. Write down every signal it sends about AI, then say on purpose what you actually mean for the next two assignments. You leave holding a finished AI card you can post today: what is allowed, what is not, what students try first, what a tool is for, and which question comes to you. You do the thinking. Nothing here writes it for you, and nothing you type leaves this page.

Why this exists. In a randomized trial at the University of Maryland, only about 15 percent of students with access to an AI tutor ever used it. In sections of the same course, final grades still came in about four points lower, and recorded participation fell sharply. The authors suspect the tool may have signaled a broader acceptance of AI, and they say they cannot establish that. Separately, in a Dartmouth survey of 1,003 U.S. college students, 71 percent agreed that an instructor’s express permission or prohibition guides how they use AI on an assignment. This session is one way to say it on purpose. It is a judgment call, not a tested fix, and it will not tell you who used AI.

SEED

Read your course the way a student does

One course. List everything a student can see or hear about AI in it right now: a tool in the course site, a line in the syllabus, something you said on day one, a campus tool someone else switched on. Then write what a reasonable student would conclude. Use their words, not yours. That sentence stays with you.






SPROUT

Say it expressly for the next two assignments

In the Dartmouth survey, an express permit or prohibit is what most students said guides them. So be express, one assignment at a time. Name the task, not the tool, because the tools keep changing. Check these lines against your syllabus and your department’s rules.












GROW

Say what comes first, and what a tool is for

At Maryland, 73.8 percent of requests to the tutor sought information, explanations or solutions. Feedback on a student’s own attempt came to 0.7 percent. Say which use you want. If no tool sits in your course, write the line for the tools your students already have. Press a button for a starter, then make it sound like you.



HARVEST

Send one kind of question to a person

In the Maryland survey, students in tutor classes were more likely to say they asked the instructor fewer questions, though few students answered that survey. An open door is easy to ignore. One named kind of question, with a place and a time, is harder to ignore.








0 of 15 fields filled

The Syllabus Generator in the Harvest Kernel Faculty Toolkit builds a student centered syllabus, and everything in it is grounded in your Course Record: your course, your outcomes, your policies and your voice, planted once and used by every tool. Full access for 14 days, cancel in one click. Write this first card by hand anyway, because this sentence has to be yours.

See the Syllabus Generator