Exam technique

The AIGP tests judgment more than recall: most questions present several defensible-sounding options and ask for the best one. Content knowledge gets you to the final two options — technique picks between them. Drill this alongside the reference material, and re-read it the week before your exam.

1 · Spot the signal words

Read the stem twice before looking at the options — the options are written to anchor you. These words change what the question is asking:

"MOST important" / "PRIMARY" / "BEST"

Several options may be valid — pick the root cause or most foundational one, not just a correct one.

"FIRST" / "BEFORE"

A sequencing question. Pick the earliest step in the lifecycle, not the most important overall.

"EXCEPT" / "NOT"

Flip your thinking: three options are correct — find the odd one out. Missing this word makes you pick the exact opposite of the right answer.

"LEAST likely" / "LEAST relevant"

Find the weakest connection. The other options will all be clearly connected.

Role words (provider, deployer, controller, processor, importer)

The answer depends on WHO. Verify the obligation belongs to that role before judging what the obligation is.

Framework words (EU AI Act, GDPR, NIST, OECD, ISO)

Match the answer to that specific framework — a real requirement from the wrong framework is a distractor.

2 · Know the distractor patterns

Wrong options aren't random — they're built from a small set of recurring patterns. If an option matches one of these, it's almost certainly a distractor:

✗ 1Absolutes

"Always", "never", "all", "eliminates", "guarantees", "solely". Governance is contextual and risk-based — absolute claims are almost always wrong.

✗ 2Responsibility shifting

"The vendor is solely responsible", "the AI acted autonomously". You can outsource tasks, never accountability.

✗ 3Single safeguard

"Anonymize the data and all concerns are resolved", "add a disclaimer". Governance is layered — one control never resolves everything.

✗ 4Right concept, wrong role

A real obligation assigned to the wrong party (e.g., a provider duty pinned on the deployer). Verify WHO before WHAT.

✗ 5Right concept, wrong phase

A valid activity in the wrong lifecycle stage — impact assessments happen pre-deployment; monitoring is post-deployment.

✗ 6Right concept, wrong framework

NYC LL144 requirements are not EU AI Act requirements; GDPR access rights are not its transparency duties.

✗ 7Wrong risk tier

Social scoring is prohibited, not high-risk. CV screening is high-risk, not prohibited. Chatbot disclosure is a transparency duty, not high-risk.

✗ 8Aggregate metrics

Overall accuracy used to justify deployment. Aggregates mask subgroup failures — fairness needs disaggregated analysis.

✗ 9Dismissing the concern

"Public data has no privacy protection", "this isn’t really a risk". Options that wave away a governance concern are almost always wrong.

✗ 10Disproportionate response

"Cancel the project entirely", "stop using AI". Governance seeks proportionate mitigation, not abandonment.

3 · Tiebreakers for the final two

4 · The per-question routine

100 questions in 180 minutes ≈ 1.8 minutes each. A repeatable routine keeps you disciplined when options look similar:

  1. ~15s Read the stem twice. Circle the signal word (MOST / FIRST / EXCEPT / role / framework). Decide what is really being asked before reading any option.
  2. ~10s Identify the lens: which domain, which framework, which role, which lifecycle phase?
  3. ~20s Eliminate 1–2 options using the trap patterns (absolutes, wrong role, wrong tier, wrong framework).
  4. ~15s Stuck between two? Apply a tiebreaker principle. Then commit and move on — ~1.8 min/question is the budget, so quick questions bank time for case studies.

5 · Question formats to expect

Beyond standard single-best-answer items, the official practice exam confirms these formats appear — each with its own discipline:

Multi-question scenarios

A long business scenario (named company, named people, often international) followed by 2–4 questions. Read the scenario ONCE carefully, noting: jurisdiction(s), each actor's role (provider/deployer…), the lifecycle phase, and any red-flag facts (special-category data, automated decisions, children). Each question probes a different angle of the same facts — don't re-read the whole scenario per question; consult your mental map. Our practice exams include scenario sets that work the same way.

"Select ALL that apply"

Unlike "pick N of M", you are NOT told how many options are correct — and there's no partial credit. Judge each option independently on its own merits (true/false against the stem), then select every "true". Don't second-guess by count.

Ordering / sequencing items

Occasionally you must put lifecycle steps in order (all-or-nothing, one point). The anchor sequence to internalise: business problem → use cases → applicable laws → necessary data → gaps & risks for planning; and plan → design/data → build/train → test (TEVV) → readiness → deploy → monitor → retire for the lifecycle. Strategy precedes law precedes data precedes risk.

Exam-day reminders

Technique framing inspired in part by the excellent free AIGP Certification Playbook by Daman David Pant — worth reading alongside this site.