In this issue:
→ The line between AI-informed and AI-decided, and why it’s blurring
→ What decision ownership actually requires now
→ One tracing practice that keeps the call yours
Here’s a test I’ve started giving executives.
Think of a significant decision you made this quarter. Now answer honestly: if AI had recommended the opposite, would you have decided differently?
If yes, ask who really made the call.
This is no longer a niche concern. The recommendation now arrives before the decision does, and the only durable advantage left is the judgment you bring to it.
Notice the order. Human first. The system where AI decides and humans ratify is easy to copy. Every competitor has it.
The subtle shift
Decision ownership erodes in stages, not moments:
→ First, AI informs the decision
→ Then it frames the decision
→ Then it effectively makes the decision, and you supply the signature
The title stays yours the whole way down.
The hidden risk
Leaders who can’t reconstruct why they decided something have already transferred the role.
This is Cognitive Atrophy at the highest-stakes level: Ownership Decline, playing out one comfortable ratification at a time.
The leadership shift
AI-informed is a strength. AI-decided is a resignation letter nobody wrote.
The difference is whether your reasoning could survive without the recommendation.
THE HANDOFF LOCATOR
Four stages. Pick one real decision from the last quarter and find where it actually sat. Two minutes.
STAGE 1 — INFORMED
You formed a view, then used AI to test it, fill gaps, or find what you’d missed.
The tell: you can state what you thought before you asked.
This is the strong position. AI made your reasoning better without replacing it.
STAGE 2 — FRAMED
You went to AI with the question and it gave you the options. You chose well among them.
The tell: your decision was a selection from a menu you didn’t write.
Still yours in a real sense. But the highest-leverage move in most decisions is deciding what the options are, and that part happened somewhere else.
STAGE 3 — LED
The recommendation arrived first. You evaluated it, agreed, and proceeded.
The tell: if the recommendation had been the opposite, you’d have needed a reason to override it rather than a reason to accept it.
The burden of proof has quietly inverted. That inversion is the whole thing.
STAGE 4 — RATIFIED
The output was the decision. Your contribution was approval.
The tell: you cannot reconstruct your reasoning independently of the output’s reasoning, because there wasn’t any.
The title is still yours. The decision isn’t.
Reading it
Stage isn’t a verdict. Stage 4 is fine for a Tier 1 decision and a problem for anything that travels. The question isn’t which stage you’re in. It’s whether the stage matches the stakes.
The direction matters more than the position. Run this on three decisions across three months. If the stages are drifting downward, that’s the finding — and it drifts by comfort, not by choice.
The uncomfortable one is Stage 3, because it feels exactly like Stage 1 from the inside. Both involve reviewing, agreeing, and proceeding. The only difference is which way the burden of proof was pointing, and nobody notices that in the moment.
Three moves this week
Trace one decision. What shaped my conclusion? Which assumptions are mine? Where did AI inform versus where did I decide?
Decide first, consult second on one meaningful call. Then compare. The gap between your answer and AI’s is where your judgment lives.
Say the ownership sentence out loud when it counts: this is my call, and here’s my reasoning.
Question of the week: Of my last five significant decisions, how many could I have defended if the AI recommendation had never existed?
Before AI reshapes your organization, make sure it isn’t reshaping your ability to think.
The Human Intelligence Assessment is free and takes about ten minutes. It measures the three capabilities AI cannot replace: Awareness, Adaptability, Authenticity.
→ Take the free Human Intelligence Assessment
Lead what AI can’t.
JoshHMiller.com