Own the Line

There’s a difference between handing AI the work and handing it the wheel. Most of the anxiety about AI and knowledge work collapses that distinction. The fear isn’t really that the model gets an answer wrong — it’s that we stop noticing whether it’s wrong at all.

That slide has a name in Fragile: the move from intellectual laziness to cognitive surrender. The first is recoverable. The second is the one to guard against.

Two tracks, two betrayals

Working with AI runs on two tracks. On the first, you offload work — drafting, summarizing, scaffolding — to move faster. On the second, you use AI to make yourself sharper: to pressure-test a position, to surface the question you didn’t think to ask.

Each track has a failure mode. Track one fails when speed quietly becomes the only goal and you stop reading what you shipped. Track two fails when the sparring partner becomes the decision-maker, and your judgment atrophies from disuse.

Use AI to think deeper, not lazier. Simulate, don’t surrender.

The line is a verb

“Owning the line” isn’t a one-time setting. It’s a practice — the deliberate act of deciding, every time, which calls are yours to make and which are safe to delegate. The line moves with the stakes: a throwaday first draft sits on one side of it; a strategic bet that shapes a quarter sits firmly on the other.

The tell that you’ve lost the line is subtle. It’s not a bad output. It’s the moment you’d struggle to explain why the output is good — when you can no longer reconstruct the reasoning, only repeat the conclusion.^[1]

What this looks like in practice

• Keep a reason, not just a result. If you can’t say why, you haven’t decided — you’ve deferred.
• Delegate the procedural, defend the consequential. The repetitive blocks are exactly what AI should carry. The load-bearing judgment is not.
• Re-earn your opinions. They’re rented, not owned; the rent is due every day.

AI is a superhuman sidekick. But a sidekick with the controls is just a pilot you’ve stopped being. Own the line.