Three Tiers
Cheap to get wrong, easy to undo. Use AI freely and move on.
Example: "Which two themes should I emphasize in this week's team meeting opening?"
Real consequences but recoverable. AI can summarize and flag; you apply organizational context and relationship knowledge before deciding.
Example: "Should I approve or defer this vendor contract renewal?"
Consequential, hard to reverse, and accountability-heavy. AI can synthesize and flag consistency, but you must own the call. No AI recommendation gets accepted without independent review.
Example: "Which candidate should I advance to the final round?"
Dana's story
Dana faces three decisions in one week. Which themes to open the team meeting with: low-stakes, reversible, so Dana brainstorms with AI and moves on in two minutes. Whether to renew a vendor contract: medium-stakes, so Dana uses AI to summarize the terms and flag risky clauses, then applies what Dana knows about the relationship and the budget before deciding. Whether to advance a particular candidate: high-stakes people decision, so Dana uses AI only to check the interview notes for consistency, then owns the call personally, documents the reasoning, and screens for bias before acting. Same tool, three completely different levels of scrutiny.
Apply It Today
Take the next three decisions on your plate and sort them into the three tiers. Match your verification effort to the tier, not to how busy you happen to be.
Reflect