LANDMARK · 7 MIN · JUDGE
Match checking to risk
After this landmark, you can sort an AI output into a green, yellow, or red zone and pick a verification effort that fits the stakes.
Checking everything to the same depth wastes effort on the trivial and under-checks the dangerous. The better question is not “is this true?” but “what does it cost me if this is wrong, and can I undo it?” Green-zone work is low-stakes and reversible — a brainstorm, a first draft, a casual explanation — where a quick sanity read is enough. Yellow-zone work feeds a real decision or goes to other people, so you confirm the load-bearing facts before you rely on them. Red-zone work is high-stakes or hard to reverse — anything legal, medical, financial, safety-related, or published under your name — where you verify every claim against a primary source, or don’t use it. The skill is placing the task in a zone honestly, before the fluent output tempts you to skip the step.
This is the canonical concept. It stays the same across learner lenses so personalization never changes the underlying facts.
◌ EVERYDAY LENS
What this looks like for you
Before acting on an answer, ask “if this is wrong, what happens?” A recipe substitution is green; a dosage, a tax rule, or a legal deadline is red — open a real source. Reversibility is your fastest test.
MAKE A DECISION
A colleague pastes an AI-drafted paragraph into a client proposal, including a specific market-size figure. What is the right check?
CARRY THISTake one AI output from today, name its zone in a sentence, and say what check that zone demands.