A SeedStacking Field Kit · Academic Integrity

The Trust Shift Kit

Replace AI detection with design in four moves. No software, no false accusations. One assignment, about an hour across the week.

Why this kit exists. AI detectors do not detect AI. They measure how predictable your sentences are, then dress that guess up as evidence. They carry a documented bias against clean writers and non native English speakers, and they cannot defend a single verdict on appeal. Universities are switching them off. This kit is what you run instead.

0 of 12 reps complete. Work one phase at a time.

Seed

Retire the detector, write the policy

5 to 15 min

Pick one assignment. Turn the detector off for it. Replace it with a single clear sentence students cannot misread.

Choose one upcoming assignment to run through the Trust Shift.
Write one sentence naming where AI is welcome and where it stays out.
Put that sentence at the top of the assignment, not buried in a syllabus.

Policy line template: On this assignment, AI is welcome for ______. AI stays out of ______. If you use it, show your work using the Trust Receipt below.

Sprout

Require the receipt

15 to 20 min

Ask for the work behind the work. The receipt makes thinking visible instead of hidden, and a student who cannot produce one is telling you something no detector could.

Add a short Trust Receipt students submit with the final draft.
Ask for the prompts they used, the output they got, and what they changed.
Make it low friction: half a page, not a confession.

Trust Receipt: 1) The prompt(s) I used. 2) What the AI gave me. 3) What I changed, cut, or corrected, and why. 4) One thing the AI got wrong.

Grow

Grade the judgment, not the typing

20 to 30 min

Build a short rubric that rewards what the student added on top of the machine. When judgment is what scores, a slick AI draft stops being a shortcut and becomes a starting line.

Write three rubric lines that only a thinking human earns.
Give real weight to catching the AI’s errors and pushing past its first answer.
Test it: could a raw AI draft score well? If yes, raise the bar.

Rubric starters: Caught and fixed a factual or logical error the AI made · Made a choice the model would not have made and defended it · Connected the work to something specific from our course.

Harvest

Let them defend it out loud

30 to 45 min plus shareable

Close the loop with a two minute checkpoint. No humanizer survives a follow up question about why the third paragraph says what it says. Then hand your one pager to a colleague.

Design a two minute defense: one oral question, or an in class paragraph.
Run the full loop on your one assignment this week.
Write a one page summary and share it with one colleague in your department.

Defense prompt: Walk me through one choice you made on this that you are proud of. Why that, and not the obvious version?

Want the how, not just the what?

The kit gives you the moves. The community gives you the reps, the examples, and educators solving the same problem.

Join the Free Community

Plant Ideas. Cultivate Skills. Harvest Results.