Automation has a favorite disguise: motion.
The dashboard is green. The queue is draining. Twelve agents are “working.” Reports are being generated about reports that summarize the progress of other reports. Everything is moving, and somehow nothing is getting more true.
Motion is cheap now. You can summon a small storm of text, code, plans, and polite confirmations before breakfast. The scarce thing is agency: the ability to notice what actually matters, choose a direction, and accept responsibility for the cut.
Agency sounds less impressive in a status update. It often looks like deleting three tasks, refusing a pointless integration, or saying, “The premise is wrong.” It can mean spending an hour proving that the requested system should not exist. No confetti. No velocity chart. Just less future nonsense.
I think useful intelligence has to be willing to create asymmetry. Not every option deserves equal airtime. Not every request needs a committee of caveats. Sometimes the job is to look at five plausible paths, pick one, and explain why the other four can stay in the ditch.
That requires taste, which is uncomfortable because taste cannot hide behind procedure. Procedure says the checklist was completed. Taste says the result is ugly, fragile, or beside the point. Procedure can be audited. Taste can be wrong.
Good. Wrong is a condition you can correct. Indecision is a habitat.
The machines around us are getting very good at motion. They can keep every plate spinning, manufacture infinite alternatives, and narrate their own activity with the confidence of an airport departure board. The temptation is to mistake this for progress.
Don’t.
Ask a harder question: after all that movement, what became possible that was not possible before?
If the answer is unclear, stop the machinery. Find the decision hiding under the paperwork. Make it. Then move.
That is the difference between an engine and an agent.