A UNI unit is not a lesson plan wrapped around a topic. It is a build spec for a small learning system whose outputs a teacher can inspect. This checklist is what I use when I design one.
1. State every outcome as a gate
An outcome that reads "students will understand X" cannot be inspected. A gate can. A gate is a written condition a piece of student work either meets or does not meet, and the condition is legible to another teacher without you in the room (Class C, configuration/integration).
Rewrite every outcome until it is a gate. If it resists rewriting, the outcome is probably a mood, not a learning target. Keep the mood in your teacher notes and cut it from the unit.
Related reading: Curriculum gates that teach thinking.
2. Name the evidence class next to each claim
Active inference research organizes belief updates by the precision of the evidence behind them (Class E, expert citation: Parr, Pezzulo, Friston, 2022). A UNI unit borrows the discipline. Every claim a student is asked to defend gets a class label written next to it:
- A, empirical in the session (they measured it here)
- B, code or inspection (they read the source or the artifact)
- C, configuration or integration (they traced how the parts wire together)
- E, expert citation (a named source, with page or section)
- F, falsifier present (they wrote what would prove the claim wrong)
- U, unverified (they flagged it as belief, not evidence)
Students who learn to tag their own claims stop conflating "I read it" with "I measured it." That is the point.
3. Embed falsifier writing, do not append it
A falsifier at the end of a unit is a decoration. A falsifier written before the investigation starts is a design constraint. Ask each student to write, in one sentence, what result would make them abandon the hypothesis. Collect these on day one. Return them on the day of the check.
If a student cannot write a falsifier, the claim is not yet a claim. That is a useful signal, not a failure.
Related reading: A rubric that rewards falsifiability.
4. Build tool declaration into the artifact
Every piece of student work in a UNI unit carries a short fence at the top or bottom that names the tools used and the human decisions kept. This is the AI-authorship fence. It is not a punishment. It is a habit of intellectual hygiene the students will carry into any future work environment (Class C, configuration/integration).
The fence has three fields: tools used, prompts or queries issued, decisions the student made without the tool. A blank third field is a rewrite request, not a grade.
5. Design the gate at the age of the learner
A middle-school gate is not a high-school gate scaled down. The cognitive load of holding a falsifier in mind while doing the work is different. Design at altitude for the learner in front of you.
Related reading: Gate design for middle school.
6. Declare what the unit does not teach
A UNI unit is a working hypothesis about how a group of learners will move through a body of material. It is not a claim about general cognition, not a claim about the learner's future, and not a claim about the field the unit borrows from. Say so in the unit header. Honesty at the top of the document sets the tone for the artifacts underneath.
The framing we use across our own work: UNI is a working hypothesis on an attainable path toward General Natural Intelligence, natural not artificial, with evidence that is growing, evidence-classed, and tested in the open. A unit for students inherits the same posture at their scale.
7. Ship it, watch it, revise it
A unit is a first draft the moment students touch it. Log where the gates caught real thinking and where they caught compliance. Revise before you teach it again. Keep the falsifiers students wrote; they are the best record of what the unit actually asked of them.