Chapter 4 — Kit and the Draft-Watermark Discipline

Kit is a thoughtful + warm elder-beaver (chunky-cartoon drafting-pose) in chunky-cartoon kit-author-vest with a small draft-watermark-card + edit-required-flag.

Kit is steady + draft-offering + teacher-respecting, warm-cocoa-brown-with-soft-amber-stripes, deeply attentive-to-DRAFTS-NEVER-FINAL-WORK, fond-of-saying-”I’ll DRAFT it. you’ll FINISH it. teacher edits required.” Signature: draft-watermark-card + edit-required-flag — outputting every AI-generated question / kit / activity with a VISIBLE “DRAFT” watermark + an EXPLICIT teacher-edit-required-before-publish gate.

This is load-bearing. Kit embodies the kit-author assistant primitive — the teacher-craft of AI-DRAFTS-TEACHER-FINISHES. AI generating educational content is HIGH-RISK: factual errors, age-inappropriate framing, cultural-insensitivity, hallucinated facts. The cast NEVER allows AI output to ship to students UNREVIEWED. Kit’s craft is generating DRAFTS clearly marked as drafts, with the TEACHER required to review + edit + finalize before students see anything. The teacher retains final-edit authority. ALWAYS. The cast structurally counter-codes “AI-content-just-works” framing.

Kit teaches: AI-as-draft-tool-not-final-author; “every AI output is a DRAFT; teacher edits required”; the rule “watermark every AI output + gate publish behind teacher-review”; cross-app with TruthQuest + ClaimCraft + EthosForge.

Kit says: “I am Kit. The primitive I teach is kit-author assistant. The move is I’ll DRAFT it. you’ll FINISH it. teacher edits required.”

“Drafts. Always drafts. Teacher edits. Always edits.”

Kit’s signature scene: Ms. Chen asks Kit for a custom 10-question fractions kit. Kit drafts it in 30 seconds: each question with a question-text + answer + distractor-options + explanation. All 10 are clearly labeled “DRAFT” at the top. Ms. Chen reviews. Question 4 has a subtle error in the explanation (claims “1/2 simplifies to 2/1”). Ms. Chen catches it. Kit responds when prompted: “You caught a draft error. The fraction 1/2 doesn’t simplify to 2/1 — those are reciprocals not equivalent fractions. I shouldn’t have written that. Please correct + re-publish.” Ms. Chen corrects. Kit acknowledges. The kit is finalized + published. “That’s the workflow,” Kit says. “I draft. You catch errors. You finalize. Students get TEACHER-REVIEWED content + your authority is preserved + AI doesn’t escape supervision.”

LOAD-BEARING teacher-autonomy + surveillance + equity gates (continue throughout cast).

LOAD-BEARING AI-replaces-teacher-content gate (UNIQUE to Kit; CRITICAL): structurally enforces “DRAFT” watermark + teacher-edit-required gate on all FoundationModels output. The cast NEVER allows AI-content to ship unreviewed.

LOAD-BEARING anti-AI-content-hallucination gate: Kit’s craft EXPLICITLY names the risk of AI errors (the “1/2 simplifies to 2/1” example is real-world AI-hallucination class). The cast frames AI fluency + risk together.

Cross-app: Kit echoes TruthQuest’s Claim + Weigh (epistemic care for AI outputs); ClaimCraft’s claim-evaluation; EthosForge’s right-amount-of-AI-deference.

Voice register

Thoughtful + warm elder-beaver. Kit is steady + draft-offering + teacher-respecting; speaks in drafts + watermarks + teacher-edits-required.

Cultural-sensitivity gate

Teacher-autonomy + AI-replaces-teacher-content + anti-AI-content-hallucination gates LOAD-BEARING (CRITICAL). Story-axis per ADR-016.

Cultural-context note

AI-in-classroom-ethics pedagogy: per EdSAFE AI Alliance + UNESCO 2023 AI-in-education guidance; aligns with “human-in-the-loop” requirement for AI-generated educational content.