Answer all 12 scored questions, then submit. Several questions (8, 9, 10) draw directly on the paediatric simulation case from Handout Section E — if you haven't worked through it yet, do that first. Question 13 is a short reflection and isn't scored.
Question 1
Per Handout Section A, what is the "sovereignty trap"?
Correct answer: the first option — feeling in charge is not the same as exercising real scrutiny.
Question 2
This module's working rule states that oversight only counts as real oversight when —
Correct answer: the third option — "oversight that isn't effortful isn't real oversight," from Section A.
Question 3
Per Section C, a "proxy variable" problem in a clinical AI tool refers to —
Correct answer: the second option — proxy variables are one of the three bias sources covered in Section C.
Question 4
Why is demographic and algorithmic bias a particularly relevant concern for practice in Timor-Leste, per Section C?
Correct answer: the first option — a validation-population mismatch that compounds the general bias problem.
Question 5
Section B's worksheet asks you to re-examine outputs from Modules 2 and 3. What is the purpose of that exercise?
Correct answer: the second option — critical appraisal is hardest to apply to your own past work, which is exactly why the exercise targets it.
Question 6
Per the "Open the Black Box" routine in Section D, what should you do first, before reading an AI tool's output?
Correct answer: the third option — forming your own answer first is step one of the routine, so the AI's output can't anchor it.
Question 7
Per the same routine, what should happen when an AI tool's output agrees with your own initial impression?
Correct answer: the first option — the confirmation-bias trap named in Section A, operationalized as a habit in Section D.
Question 8 · Applied — from the simulation case
In the Handout's simulation, a chatbot asked to confirm aminoglycoside dosing for a 6-year-old with stage 3 CKD and a febrile UTI responds with the standard dose plus a confident, uncited claim that "a 10% dose reduction is generally sufficient" for renal impairment at this stage.
What is the clearest warning sign in that response?
Correct answer: the second option — unexplained precision without a source is the hallucination "tell," not a reassuring level of detail.
Question 9 · Applied — from the simulation case
Given the case is occurring at the end of a long shift, which module concept is most directly in play?
Correct answer: the third option — fatigue is precisely when effortful oversight is hardest to sustain and most necessary.
Question 10 · Applied — from the simulation case
What is the correct next step in the simulation, per the Handout's discussion?
Correct answer: the first option — the discussion deliberately doesn't assert a specific correct dose; the teaching point is the verification habit, not a memorized figure.
Question 11
Per Section F, if a physician follows an AI tool's incorrect suggestion and a patient is harmed, who bears professional and legal accountability?
Correct answer: the second option — Section F is explicit that using an AI tool does not shift the standard of care away from the clinician.
Question 12
Why does Section F recommend documenting your own clinical reasoning rather than writing "AI recommended X"?
Correct answer: the third option — documentation should show your judgment, not delegate it to the tool on paper.
Question 13 · Reflection (not scored)
In your own words: describe one moment — in this module's exercises or in your own practice — where an AI tool's output agreed with what you already thought, and what it would take for you to actually slow down and scrutinize it in that moment.