By Caleb Barnette
If you've spent any time around the AI conversation in the past two years, you've probably been handed two takes that don't have much room between them.
The first: AI is going to replace most of what you do. Your job, your team, your industry. Don't get left behind.
The second: AI is overhyped. You've tried it, it gets things wrong, it's a fancy autocomplete. None of this is real for businesses like yours.
Both takes treat AI as a thing that either does the work or doesn't. Replace, or stay irrelevant. That's a false binary, and it's costing people on both sides of it.
The truth is in the third option, which is the one almost no one is selling: AI doesn't replace your judgment. Used well, it accelerates what your judgment can do.
That sentence isn't a hedge. It's a different way of thinking about what AI is for.
First, the slop
Anyone who's spent ten minutes with AI on a real task has seen it produce something that, in the most accurate technical term, is slop.
Slop is what AI makes when no one is paying attention. It looks fine on first read. Three sentences in, you start to notice the words aren't doing any work.
You've seen it. You've maybe even shipped some of it.
Open Zillow and pull up the last twenty new listings in your county. Half of them have been written by AI now, and most of those reads like this:
"Welcome home to this stunning, well-appointed retreat that perfectly blends modern comfort with timeless charm. Whether you're entertaining guests or relaxing after a long day, this property offers the ideal blend of style and functionality."
Read it twice. It's saying nothing. There's not a single fact about this house in those two sentences. A buyer would learn more from one photograph than from the entire description.
Or look at the "About" page on the website of any contractor who's outsourced their copy to AI in the last year:
"At [Company Name], we're passionate about delivering exceptional results for our valued clients. Our team of experienced professionals is committed to providing innovative solutions tailored to your unique needs."
Or the customer-service email that's clearly been AI-drafted and not edited:
"Thank you for reaching out! We truly understand how frustrating this situation must be, and we want to assure you that your concerns are very important to us. We're taking your feedback into careful consideration as we continuously strive to improve."
In all three cases, the AI did exactly what it was asked to do. It produced text. The problem isn't that the AI is broken. The problem is that nobody on the human side did the work that turns text into communication.
And the slop has a cost. A buyer scrolls past the listing because nothing in it told them what's special about the house. A potential client clicks off the contractor's website because every word reads like every other contractor's website. A customer's complaint gets worse, not better, because the response felt like a form letter from someone who didn't read it. The AI didn't lose the deal, the lead, or the customer. The decision to ship the AI's first draft did.
That work isn't optional. It's the thing.
What AI is actually for
Here's the boring truth: most of what makes AI useful is the work you do before and after it runs.
Before, that work is knowing what good looks like. Knowing what specifically you want. Knowing the customer, the property, the case, the situation. Knowing the parts of your business that are worth saying out loud and the parts that aren't.
After, that work is reading what came back. Catching what's wrong. Catching what's missing. Catching what sounds right but isn't. Cutting the empty sentences. Adding the specific detail the AI couldn't possibly have known.
The AI does the part in the middle. The fast part. The part where you have a draft suddenly, instead of a blank screen.
The before and after is the part the human keeps. That's the part that turns a tool into something useful.
When we describe this to clients, we usually put it in three pieces.
Three jobs the human keeps
1. Taste. Does this sound right? Does it sound like us? Does it sound like a person who knows what they're doing?
AI cannot answer this. AI averages. It generates the most predictable continuation of whatever you've given it. Predictable is the enemy of good copy, good strategy, and good anything that's supposed to come from a specific person, in a specific place, with a specific point of view.
Taste is the thing that catches "welcome home to this stunning, well-appointed retreat" and replaces it with "three bedrooms upstairs, a kitchen redone in the last five years, and a back porch that catches the afternoon light from May through September." The first one was generated. The second one was observed.
2. Truth. Is this right? Did the AI just make something up? Did it confidently state a number, a date, a precedent that doesn't exist?
AI hallucinates a meaningful percentage of the time on tasks it isn't equipped for. "Hallucinate" is the technical word; in plain English, it makes things up. And it does this with a confident voice. The voice does not change when the AI starts inventing. There is no warning label.
The human's job is to know enough about the topic to spot the false confidence. A real estate agent knows what a 1972 ranch in their county actually goes for. A bookkeeper knows what depreciation schedules are legal. A general contractor knows which permits the county will and won't accept. The AI doesn't. The AI is averaging.
3. Trust. Is this something you'd put your name on?
This is the one that gets skipped most. People treat AI output the way they'd treat a draft from a junior employee, something to clean up. But a draft from a junior employee is something you trained, something whose mistakes you understand, something that gets better the more you work together.
AI doesn't get better the more you work with it on a single task. It gives you a different version of the same average. The judgment about what to ship is still yours. The reputation that goes out the door is still yours.
If you wouldn't sign your name to it, don't ship it. That's not a high bar. Most people clear it less often than they think, especially when AI made the work feel free.
What this looks like for a Charleston business
Most of the small businesses we work with around Charleston aren't trying to replace anyone with AI. They're trying to figure out whether it's worth bringing in at all, and where.
The framework above is the answer.
If a task is mostly about taste, truth, or trust, the things only you could write or decide, AI is bad at it on its own. But it can be useful next to you, drafting the first version, taking the blank-page friction off the day. The output is still yours. The acceleration is real.
If a task has none of those three properties, AI can run it almost end to end: summarizing meeting notes, formatting a spreadsheet, transcribing a call, cleaning up a CSV, drafting a routine confirmation email. Those are the wins that pile up quietly. Most clients we work with are surprised, in the first month, by how many of these tasks they were doing by hand.
The trap is in the middle: tasks that look mechanical but actually require taste, truth, or trust. Real estate listings look mechanical. They aren't. Customer-service emails look mechanical. They aren't. The "About" page on your website looks mechanical. It very much isn't.
That middle is where the slop comes from. Not because the AI is broken, but because the human stopped paying attention at the moment they should have started.
The multiplier
The conversation about AI keeps wanting to be about replacement. It's the cleaner story. Either it takes your job or it doesn't matter to your job, and we can move on to the next thing to argue about.
But replacement isn't what's actually happening, and irrelevance isn't either.
What's actually happening is that AI is putting a multiplier in front of every person who chooses to use one. The multiplier acts on whatever the person already brings. Skill, attention, judgment, taste: all multiplied. Carelessness, sloppiness, confidence without competence: also multiplied.
The slop you've been seeing on Zillow, in your inbox, on the websites of competitors who let go of their copywriter is the multiplier acting on something that wasn't worth multiplying in the first place.
The acceleration you've been hearing about, the version that's quiet but real, is the multiplier acting on someone who knew what they were doing before the multiplier showed up.
You don't get to opt out of the multiplier. It's already here, and it's already changing what your competitors can produce in a day.
What you do get to choose is what it multiplies.