You've seen the framing. Half the AI commentariat has settled into a shorthand where the model is the brilliant one and the human is the operator — the person flipping the switch, the one who uploaded the doc and waited for the magic. There are companies named after the metaphor. There are conferences whose entire premise is teaching you to be a good one.
I'm here to argue the metaphor is wrong, the word is doing damage, and there's a better one already sitting in the dictionary.
What "operator" actually says
An operator runs a machine that does something more impressive than they could. The crane operator does not lift the I-beam — the crane does. The operator's job is to position, lever, brake, and stay out of the load's way. The skill is real. The ceiling is the machine.
That metaphor used to describe the human in the loop is doing three things at once, and none of them are good.
- It locates the intelligence in the wrong place. The model is a flexible, language-shaped tool. The judgment, the taste, the sequence — that's still you. If the model is the smart one, then your job is to hand it a steering wheel and pray. That's not what the work is.
- It teaches the wrong skill. If you believe you're an operator, you'll over-invest in tool stacks and under-invest in taste. You'll collect prompts the way some people collect knives — sharp, expensive, and almost always still in the drawer. Tools without judgment is a kitchen full of equipment and nothing to eat.
- It frames you as replaceable. An operator is, by definition, the bottleneck the machine is trying to remove. If your story for what you do is "I operate AI," you have already conceded the argument that someone with a faster wrist could do it better. They couldn't. But you've conceded the argument anyway.
What the work actually is
The work is not pulling levers. The work is asking the right thing of the machine, in the right way, at the right moment in the project. The work is reading the output and knowing — not guessing — which 20% of it is worth keeping. The work is steering the conversation across five turns until the model produces something you couldn't have written, and you can tell, because you've been writing this for twenty years.
That's not operating. That's practicing.
The word I'd use instead
I'd use practitioner. Or, depending on the context: working witch, prompt engineer, artist, artisan, maker, builder, writer-in-chief. All of these put the agency where it belongs — on the human — and treat the model as the instrument, not the genius.
"Practitioner" works because it carries the right shape: someone who is practicing something — building skill over time, by doing the thing repeatedly, with attention. It is the word we use for the people in the world we trust most. We have medical practitioners. We have practicing attorneys. We have spiritual practitioners. We do not have medical operators. There is a reason.
The witches in old stories were the women who paid attention. They knew which plants did which things because they had time and curiosity and a willingness to be wrong. They were practitioners. The "magic" was a thousand small observations stacked on top of each other until the stack started to lean toward the truth.
That is the metaphor I want for working with AI. Not the operator at the controls. The practitioner with a workbench, paying attention, writing it down.
What changes when you swap the word
Try it for a week. Replace "I'm an AI operator" with "I'm a practitioner." Replace "operating model" with "practice." Replace "tool stack" with "workbench." Watch what happens.
You'll start asking different questions. Instead of "what's the best tool?" you'll start asking "what am I trying to make, and what's the smallest workbench that could let me make it?" Instead of "how do I get faster?" you'll start asking "what part of this do I want to stay slow on, on purpose, because it's where the work earns its keep?"
You'll notice that the people whose work you actually admire — the writers, the strategists, the founders, the consultants who keep landing real outcomes for real clients — already use the practitioner framing without naming it. They don't talk about their stack. They talk about their craft.
What this means for A&I
This is why the language on this site never calls you an operator. The dogs are operators. Lucy is the Chief Bark Officer. Linus is the Senior Bark Analyst. They are literally standing by. They flip exactly the kind of switches an operator flips, which is to say none.
The humans? You're a practitioner. Or you're becoming one. Either way, that's who this lab is for.
— L.