THE SHORTCUT LAB — COPY KIT
Kalahari 2026

KEY QUESTION
What did this task require the student to do that mattered?

FIVE CAPACITIES
Memory — internal material thinking can work with.
Judgment — deciding what is true, useful, ethical, relevant, and worth doing.
Dialogue — making thinking public through listening, pushback, and revision.
Productive Struggle — purposeful friction where skill and durable understanding form.
Agency & Authorship — the learner's ownership of thought, choice, revision, and action.
Destination: Authentic Understanding

ASSIGNMENT AUDIT
1. THE ARTIFACT. What does the student actually submit—the thing that lands in the gradebook?
2. THE HIDDEN WORK. What thinking is that artifact supposed to prove happened?
3. THE SHORTCUT TEST. If a student pasted the directions into an AI tool, how much of a full-credit submission might come back? 0% / 25% / 50% / 75% / 100%
4. THE COST. If AI did those parts, which capacity might the student never build?
5. THE EVIDENCE GAP. What part of the process is visible today, and what one piece of human work must the task still force into existence?

REDESIGN PROMPT
Pick one capacity. Rewrite one component. Use the smallest change that makes the human work unavoidable.

MOVES MENU
MEMORY
- Retrieval before resources open
- Explain from memory in 60 seconds
- Re-explain the idea two weeks later

JUDGMENT
- Rule on an AI answer: find the flaw or verify the claim
- Choose and defend a position
- Make a credibility ruling with justification

DIALOGUE
- Move the assessable moment into live conversation
- Use a 30-second voice reply with required uptake
- Add a three-minute defend-your-post conference

PRODUCTIVE STRUGGLE
- Stage proposal, attempt, feedback, and revision
- Grade the change between draft one and draft two
- Analyze an error from the learner's own attempt

AGENCY & AUTHORSHIP
- Anchor the task in data the learner gathered
- Give a meaningful choice of angle or format
- Add an author's note: what I changed and why

RED-TEAM
How would a shortcut-seeking student still AI their way through this? Patch one hole without turning the task into surveillance.

WELLNESS GATE
If your redesign doubles your grading, redesign the redesign.

BOUNDARY TEMPLATE
For this task, AI is [● GREEN/OPEN / ▲ YELLOW/ALLOWED WITH RECEIPTS / ⛔ RED/CLOSED] for [this part of the task], because the learning you're building here is [the human work, in student language]. [If allowed: name the receipt. If closed: explain why this part needs to come from the learner.]

Shape + label always. Do not communicate the boundary with color alone.
"Open" and "allowed" always mean the tools your school has vetted and approved for that purpose — if no approved tool exists for that part, it stays closed until one does. And open still means say-what-you-used.