The problem with AIs isn’t what they make. It’s what they ask us to see. The old jobs—writing, drawing, composing—were simpler. You sat down, did the thing, sent it out. Now the job is staring at the screen, sifting through endless drafts, trying to figure out which version doesn’t sound like a dishwasher manual. The real work isn’t making content. It’s deciding what’s worth keeping.
A few special people used to pride themselves on knowing how to put things together. Not just words on a page, but the spaces between them. They could tell when a sentence was off, when a chord didn’t fit, when a color made your eyes ache. Now we’ve got tools that spit out 1000 options in the time it takes to brew coffee. The trick isn’t using the tool. It’s knowing which result matters. Most would rather hit “generate” again than sit quietly, squinting at the screen, asking, Is this right?
The smart future depends on the picky. Not the prompt engineers or the algorithm whisperers. The ones who can look at a paragraph and spot the lie in the third sentence. The ones who hear the synthetic twang in a melody meant to sound human. The ones who notice the shadow that doesn’t match the light. These aren’t technical skills. They’re habits. The kind you lose if you don’t use them.
One of my mentors shot to stardom by writing obituaries. He’d sit at his desk, red pen in hand, crossing out clichés. “Loving father,” “devoted husband,” “passed suddenly”—all got the slash. “Be specific,” he’d mutter to himself. “Tell me something true.” The job wasn’t writing. It was pruning. Cutting back the noise until what remained felt lived-in. Now imagine poor David facing a stack of AI-generated eulogies. Same task, but the pile’s a mile high. How long before he starts letting the clichés slide?
That’s the trouble. The machine doesn’t care. It’ll give you competent, lifeless prose forever. The burden shifts to the person in the chair. Can they still recognize warmth? Authenticity? The faint pulse of a real idea? Or have they been numbed by the sheer volume of almost good enough?
The worst part? People hate this work. Always have. Give them a choice between creating and critiquing, they’ll pick creating every time. Writing a first draft feels like progress. Editing feels like doubt. But the draft’s the easy bit. The doubt’s the point, and we know why: second thoughts are doubly hard to come by when you’re not even capable of first thoughts.
Take music. Used to be, you’d hum a tune, scribble notes, play it on an out-of-tune piano. Now you type “sad jazz for rainy nights” into a Riffusion prompt. Out comes a symphony. Except the saxophone’s too loud in the second movement. The drums sound like they’re underwater. Fixing it means listening, really listening, to each layer. Most people would rather roll the dice again. Keep generating until something feels right. But feeling isn’t the same as knowing.
Or the designer. The AI can mock up 100 logos by noon. Reasonably clean, fairly sharp, mostly balanced. But do any of them mean anything? Does the color scheme whisper “trustworthy” or scream “generic bank”? Why is the kerning so messed up? The answer’s not in the options. It’s in the person clicking through them. If they’ve spent years skimming instead of staring, they’ll miss the difference.
We used to call this taste. An old-fashioned word. It implied time, patience, a willingness to be wrong. You developed it by reading bad books, hearing bad songs, seeing bad art. You learned what not to do. Now the machines flood the zone with mediocrity. The only way through is to become ruthlessly, stubbornly selective.
But here’s the rub: discernment is hard as hell. It’s easier to let the algorithm decide. To accept the second draft simply because the first was worse. To ship the video with the awkward cut because reshooting it would take hours. The machines enable a kind of lazy perfection—nothing great, nothing terrible, just an endless parade of “closer to fine” slop.
I think about the previous generation’s librarians. The good ones could find you a book you didn’t know you needed. They’d watch you browse, notice where you lingered, then hand you something off-script. “Try this,” they’d say. No algorithm can do that. It takes a human who’s paid attention, who’s built a mental map of the stacks. AI can generate reading lists, but it can’t wander the aisles, pulling titles on a hunch. That requires a mind trained to notice.
The nightmare isn’t robots taking over. It’s humans giving up, which most of us are dying to do. Imagine a world where every ad is written by AI, read by AI, clicked on by AI. Products designed for bots, sold by bots, bought by bots. A closed loop of nothing. No one creating, no one judging—just machines passing data back and forth like kids trading baseball cards.
You’ve seen the early signs. Websites bloated with SEO slop. Social feeds full of comments no one wrote and gooey, creepy pictures no one drew. Music playlists engineered to sound “viral.” It’s not evil. It’s indifferent. The machines don’t aim low. They aim for mediocrity. Without someone pushing back, demanding better, below average becomes the ceiling.
My mom’s uncle ran a print shop in the Mon Valley. He’d stand over the presses, checking each batch. If a color ran too light, he’d stop the machine. “Can’t sell crap,” he’d say. It wasn’t pride. It was good business sense. His mom-and-pop customers expected quality. Now imagine him today, staring at a screen full of AI-generated flyers. The colors are technically correct. The text is error-free. But the layout feels off. Does he waste an hour tweaking it? Or let it go, knowing most people won’t look twice?
That’s the choice. Every minute spent refining is a minute not spent generating more. The temptation is to move fast, ship often, fix it later. But later never comes. The content pile grows, and buried in the heap are 1000 half-choices, compromises made in the name of speed.
The fix isn’t better AI. It’s better humans. Training ourselves to slow down, to question, to reject the easy answer. To read like editors, listen like composers, look like artists. To treat every piece of machine-made content as a draft—flawed, promising, begging for a red pen.
This isn’t romantic. It’s work. The kind that strains your eyes and fries your nerves. But someone’s got to do it. Otherwise, we’re just janitors in a factory nobody visits, sweeping up digital dust.
The irony? The machines could help. Imagine an AI that forces you to justify every edit. That asks, “Why change this word?” or “What’s wrong with that image?” A tool that trains you to be critical, not just productive. But that’s not what sells. The market wants faster, cheaper, more.
So we’re left with a choice: Use the tools to make garbage, or use them to sharpen our minds. The first option pays the bills. The second keeps us human.
I miss my grandfather’s coal stove. Not the heat—the work. You had to tend it, feed it, adjust the dampers. Let it die and the room went cold. Let the AI run unchecked and the mind goes the same way.
We don’t need more content. We need more people willing to sit in the quiet, read the damn thing, and say, “Try again.”