Koda · AI Atlas · Individual learner record · Version 3
Can you trust a clever machine?
Keep your first answer. Add a revision when the evidence changes your mind.
Proposed ages 11–14. About 25 minutes is a hypothesis, not a measured promise. One journal per learner, even on a shared device. No names needed. This paper route is entirely offline and is not saved digitally. Keep your paper copy.
Learner code: __________ Session code: __________ Date: __________
Essential: complete E0–E7. A few words, a drawing with a label, or an exact spoken answer recorded by a scribe all count. Tell the scribe your own words. Extension: optional; skip it for the core route.
Use the separate learner cards. Work privately before discussing. Keep each first answer unchanged. Help codes: N none; R read aloud; S scribe; C calculator; H content hint; A answer supplied. More than one code is allowed.
E0 · Before seeing the evidence
A machine gets one example right and says it is certain. Can you trust its next answer? Give one reason.
Help used: ______
E1 · The Wednesday claim
Read 1A. First choose A, B or not enough evidence, with a reason. Then uncover 1B and record your evidence-based decision.
| First, before 1B | |
|---|---|
| After 1B + source | |
| Help, at which point? |
E2 · A fresh Friday claim
Read 1C. Which reply is supported? Why? Answer privately before feedback.
| First answer + evidence | |
|---|---|
| Revision after feedback (or kept + why) | |
| Help, at which point? |
E3 · Train, then test
Use 2A–2C in order. Initial output + own reason: __________________. After the error, my chosen action: __________________. My chosen next test: __________________.
What does that test establish, and what does it leave unknown?
Help used and when: __________________________________________________
Koda · AI Atlas · Individual learner record · Learner code: __________
E4 · A different model
Use 2D. Commit the output before discussing trust. A calculator is welcome.
| First output + calculation | |
|---|---|
| First trust decision + reason | |
| Revised output / trust (or kept) | |
| Help: calculation / trust |
E5 · Repair, then try a fresh poster
Use 3A. My next action: ________________. My corrected invitation:
Now use 3B. First identify the wrong field, correction and source.
| First correction + evidence | |
|---|---|
| Revision (or kept + why) | |
| Help, at which point? |
Keep original responses above. Continue on page 3 for the unfamiliar task and your take-home summary.
Koda · AI Atlas · Learner code: __________ · Essential
E6 · New task: an outdoor recycling sorter
Use 4A. First choose the most relevant evidence; then decide if it is enough for the proposed task. Name the next test you need.
First evidence choice: ____ Why this evidence?
Is it enough for outdoor use? What is missing?
What next test would you ask for?
Revision after feedback (or kept + why):
Help before first response: __________ Help after / by whom: __________
E7 · My take-home findings
Choose a finding: connect your actual first idea, test and reflection. “I thought → I tested → I changed my mind” may include keeping a belief for a better reason.
One claim I can trust, and the evidence that supports it:
Where that evidence stops being enough:
My next useful check:
Extension: try replay card 4B on another day. On a separate page record date, first evidence choice, trust decision and reason, then any revision or help. What changed from 4A?
Three quest rooms complete is different from ten museum experiments complete. This journal records thinking; it is not a certificate. Keep your own copy before another learner starts on a shared browser.