If AI Is Replacing You, You're Doing It Wrong

This past Sunday at church, a woman stood up to share a few words. Nothing unusual about that — people speak from the heart in that room every week. But within the first two sentences, something felt off. The cadence was too even. The transitions were too clean. Every paragraph opened with a tidy signpost and closed with a neat little bow. She wasn't speaking to us; she was reading to us. And what she was reading had, unmistakably, been written by ChatGPT.

I want to be careful here, because this isn't a story about catching someone out. She clearly cared. She wanted to honor the moment and probably worried that her own words wouldn't be good enough. So she asked a machine for help, and the machine gave her something polished. The problem wasn't that she used AI. The problem was what she handed over to it.

She didn't outsource the typing. She outsourced the testimony. And the room could tell — because the one thing a congregation actually shows up for is the sound of a real person finding real words for something that matters to them. That's the part no model can stand in for. The moment it's gone, the polish doesn't help. It hurts.

I sat there and thought about a line I keep coming back to whenever the "AI is going to replace us" conversation starts:

If AI is replacing you, you're doing it wrong.

I don't mean that as a threat or a productivity slogan. I mean it almost literally. If the thing AI produced can fully replace what you would have produced — your judgment, your voice, your fingerprints — then you weren't using it as a tool. You were using it as a substitute. And the work that comes out the other end carries exactly that: the unmistakable flatness of something nobody was really behind.

Superpowers, Not Substitutes

I've spent my career on the enterprise side of this — architecting AI into operations that move thousands of field technicians, millions of work orders, real consequences. And the principle I've watched hold true at industrial scale is the same one that played out in that church pew on a Sunday morning.

AI is at its best when it amplifies a human who knows what they're doing. It's at its worst when it's asked to be the human who should have been doing it.

Think about the difference in what the woman could have done. She could have written three honest, halting sentences of her own and asked AI to help her tighten them — keeping every bit of her meaning, just sharpening the edges. That's a superpower. Instead, she asked it for the meaning itself. That's a substitution. Same tool. Opposite outcome. The line between them is simply this: did you bring the substance, or did you ask the machine to manufacture it for you?

A copilot makes a capable pilot more capable. It does not make a pilot out of someone who never learned to fly — it just hides, for a while, the fact that nobody is flying.

We've Lived Through This Before

If this anxiety feels new, it isn't. We've run this exact experiment at the scale of entire economies. When the Industrial Revolution brought powered looms into textile mills and mechanical reapers into the fields, the fear was identical to today's: the machine is here to replace the worker. And in a narrow sense, it did replace something — the raw, repetitive muscle of the task. A single mechanized harvester could do what once took a line of laborers a full day.

But the people didn't vanish. The ones who thrived were the ones who climbed up a level: from swinging the scythe to running the machine that swings it, from spinning thread by hand to operating, maintaining, and eventually engineering the looms. The work didn't disappear — it moved. What got automated was the part of the job that was never really the point, and what remained for humans was the judgment, the oversight, the craft of running the machine well.

The farmhand who refused to touch the new equipment was the one who got left behind. Not because the machine was smarter than him — it wasn't — but because he insisted on competing with it at the one thing it was built to beat him at, instead of learning to wield it.

Every wave of automation has paid the same people back: those who learned to operate the machine, not those who tried to out-muscle it.

AI is the same shift, aimed at a different layer. The steam engine automated our backs; AI automates a slice of our cognitive busywork — the drafting, the summarizing, the first-pass synthesis. The lesson the mill workers learned the hard way is the one in front of us now: the goal was never to do the machine's job by hand, and it was never to let the machine do your job either. It was to become the person who operates it skillfully — to move up a level, where the human judgment lives.

That woman at church had the relationship backwards. She didn't step up to operate the machine; she let the machine step into her place at the podium. It's the modern equivalent of standing in the field while the harvester does its thing — except she handed over the one thing the machine was never supposed to take: her own voice.

The Test I Use

Whenever I'm about to lean on AI for something — a strategy doc, a client recommendation, a talk, an email that actually matters — I run a quick gut check. It's the same test I'd offer that woman, and it's the same test I bring into boardrooms.

Ask yourself: if I removed myself from this entirely, would the output be roughly the same? If the answer is yes, stop. You're not using AI to get superpowers. You're using it to disappear. The fix is never to use less AI — it's to put more of yourself back into the part that only you can supply, and let the tool carry the rest.

That reframing changes what you delegate:

• Delegate the friction, keep the judgment. Let AI draft, summarize, restructure, and research. Don't let it decide what you actually believe or what you're trying to say.
• Delegate the first 70%, own the last 30%. The last 30% — the specific example only you have, the conviction, the point of view — is the entire reason anyone wanted to hear from you and not from a search box.
• Delegate the polish, never the presence. A speech, a recommendation, a leadership message — the value is the person standing behind it. Outsource that and you've outsourced the only thing that was ever yours.

Why This Matters Beyond the Pew

It's easy to read this as a story about one nervous speaker. But I see the exact same pattern inside organizations every week — teams using AI to generate strategies nobody owns, reports nobody reads, analyses nobody can defend when questioned. The technology isn't failing them. They've simply asked it to replace the thinking instead of accelerate it, and the result has that same telltale flatness: technically fine, fundamentally hollow.

The people and the companies that will pull ahead aren't the ones who use AI for the most. They're the ones who stay unmistakably present in their work while letting AI carry everything around it. They show up smarter, faster, and more prepared — but they still show up. The superpower is additive. It was never meant to be a replacement, and the moment you treat it as one, everyone around you can feel the difference.

AI should give you superpowers — not think and speak in your place.