Where humans must stay loud
What you'll be able to do
- Name the liability zones: hiring, reviews, sensitive communications, and legal-adjacent or money work
- Apply a "would I sign / could it hurt someone / is it binding" gate to any task
- Keep AI as drafting support and never let it become the decider
Some calls are yours by law and ethics
Hiring and firing, discipline, evaluations, and anything binding are decisions a human must own. AI can organize the information and draft the words, but it cannot make the call and it cannot be accountable for it. Accountability does not transfer to a tool.
Hiring and people decisions
AI can draft a job post or a rejection note. It must not score candidates, rank people, or make the hire. The moment a person's livelihood is on the line, the judgment is yours and stays yours.
Reviews and evaluations
You can have AI draft a review from evidence you provide. It must never invent evidence, and it must never be the one judging the person. The facts come from you. The judgment comes from you. AI only helps with the wording.
Sensitive and emotional communications
Layoffs, condolences, disputes, hard news. AI can give you structure and a starting shape, but the tone has to be owned by a human who understands the relationship and the moment. A perfectly worded message in the wrong tone does real damage.
Legal-adjacent and money
Contracts, change orders, bids, anything with dollars or terms. AI can draft the narrative around the numbers, but a human verifies every figure and every term. The numbers are not the model's to generate, and the signature is not the model's to give.
The three-question gate
Before you hand a task mostly to AI, ask three things. Would I sign my name to this and be accountable? Could getting it wrong hurt a person or expose me legally or financially? Is any part of it binding? Any yes means the human stays loud and AI stays in a support role.