It is a question that makes most of us flinch. We like to think of our taste as a fortress-a private, sacred space built from the books we read, the films we love, and the specific way the light hit the floor in our childhood homes. We believe we choose our tools. We believe the tools work for us.
But after a morning spent sneezing seven times in a row-the kind of fit that leaves your eyes red and your thoughts fragmented-I sat down to look at my own archives and realized I have been lying to myself.
I am a digital citizenship teacher. My job, quite literally, is to tell students how to maintain their agency in a world of algorithms. I tell them to watch for the nudge, to spot the bias, and to keep their hands on the wheel. Yet, for three years, I have been using the same set of upscalers and enhancement tools to “fix” my old family photos and my professional headshots.
The Teacher’s Uncomfortable Realization
I was wrong. I thought I was using these tools to recover the truth of a blurry moment. I thought I was just “adding back” what the lens had missed. But as I looked at a photo of my mother from , processed through a high-end reconstruction engine, I realized the woman in the photo had the skin of a person who didn’t exist in .
She had the skin of a high-resolution render from . The machine hadn’t found my mother; it had replaced her with its own idea of what a “good” face looks like. This is the quietest theft of the modern age. We are not just outsourcing the labor of editing; we are outsourcing the very definition of beauty.
Yara, a photographer I know who specializes in real estate, found herself in the middle of this crisis last month. She had been using a specific AI upscaler for over a year. It was fast-it took about to turn a grainy wide-shot of a kitchen into a sharp, 4K marketing asset. It was efficient. It was “better.”
But then she went out to shoot a new listing in a historic district-a house with peeling paint and hand-carved wood that had aged into a dark, complex patina. While she stood in the living room, she found herself tilting her camera and adjusting her lights not to capture the wood as it was, but to make sure the AI would “read” it correctly later.
She was shooting for the machine. She knew that if the shadows were too deep, the upscaler would turn the wood grain into a muddy, plastic smear. If the highlights were too bright, it would reconstruct the texture as polished chrome.
Yara and the Real Estate Crisis
She had stopped being a photographer and started being a data-gatherer for an aesthetic she hadn’t even chosen. Her eye had been recalibrated. She no longer saw the room; she saw the potential output of the software.
This happens because tools are never neutral. Every piece of software carries the breath of its creators. When a developer builds an AI to enhance an image, they have to tell the machine what “sharp” means. They have to give it a “loss function”-a mathematical way of saying “this result is good” and “this result is bad.”
If the training data is full of sleek, modern offices and poreless skin, the machine will learn that “good” means sleek and poreless. When you run a
through a system designed to reconstruct detail, you aren’t just stretching pixels. You are asking a set of weights and biases to make a guess.
Because we are human, and we crave the hit of dopamine that comes with “clarity,” we start to prefer the guess over the reality. We start to think the raw photo looks “wrong” because it is messy, soft, and human. The danger isn’t that the tools are bad. Many of them, like AI Photo Master, are actually quite brilliant at what they do.
They solve the very real problem of low resolution in a world that demands 4K. They turn a task into a professional result without the $200 price tag of a human retoucher. The danger is the “default.”
I see this in my classroom every day. My students don’t want to see a photo with film grain. They think it’s a mistake. They don’t want to see a shadow that doesn’t have “recovered detail.” To them, the “real” world looks slightly broken because it hasn’t been upscaled yet.
Aesthetic Homogenization and the Coffee Shop Effect
They have lived their entire lives inside the “upscaler aesthetic”-a world where every edge is sharp, every texture is predictable, and every color is optimized for a backlit screen. We are entering an era of “aesthetic homogenization.”
It is the same reason why every new coffee shop in every city from London to Tokyo starts to look the same-the white tiles, the hanging plants, the Edison bulbs. We have an algorithm for “good coffee shop,” and we follow it. Now, we have an algorithm for “good photo,” and we are letting it rewrite our visual history.
Consider the way these tools handle texture. A person’s face is a map of their life. It has scars, uneven pores, dry patches, and the fine lines earned from laughing or frowning. Traditional upscaling-the old way-would just make those things bigger and blurrier. But AI reconstruction looks at those “flaws” and sees “noise.”
It tries to resolve that noise into a pattern it recognizes. Often, it decides that your skin should look like a blend of silk and marble. At first, you love it. You look younger. You look “clearer.” But after a hundred photos, you start to look at yourself in the mirror and feel a strange sense of disappointment.
The mirror doesn’t have a reconstruction engine. The mirror is stuck in low-resolution reality. I remember a specific moment during a lecture I was giving on deepfakes. I was showing a slide of a “perfect” AI-generated human. I pointed out the lack of micro-expressions, the way the light hit the eyes with a mathematical precision that felt slightly “off.”
“But Ms. Jade, what if the ‘off’ version is the one we like more?”
– Leo, student
That hit me harder than the seven sneezes did this morning. If we train our eyes to prefer the machine’s output, then the machine isn’t the one that’s “off”-we are. We are the ones who have drifted away from the dock.
The Neutrality Myth of Mathematical Weights
When we use a tool like an AI upscaler, we are participating in a trade. We trade the specific, messy truth of a moment for a sharp, usable, and professional-looking asset. For a real estate agent trying to sell a house, or a small business owner trying to make their product look good on a website, this trade is worth it.
It is a tool of survival in a visual economy. You need that 4K clarity to compete. You need the turnaround time because you have a business to run. But we must recognize that this is a trade. If we don’t, we lose the ability to appreciate the “un-enhanced.”
The Truth Folder: Valuing the Blur
We lose the taste for the grain. We lose the value of the blur. I’ve started forcing myself to keep the “bad” versions of my photos in a separate folder. I call it the “Truth Folder.” It’s full of photos that are slightly out of focus, photos where the light is weird, photos where my mother’s skin looks like skin.
I do this to remind myself what the world actually looks like. We think we are the masters of our digital tools. We think we are just clicking a button to “improve” an image. But every time we accept a default, we are letting a developer in a dark room somewhere else decide what our memories should look like.
The path forward isn’t to stop using the tools. That’s impossible and, frankly, unhelpful. The power of reconstruction is a gift for anyone who has ever lost a precious photo to a bad camera or a low-res save. The path forward is to use the tools with a sense of “aesthetic defiance.”
Use the upscaler. Get your 4K output. Save the time. But don’t let the machine’s “idea” of a good photo be the only one you carry in your head. When you look at the result, ask yourself: What did the machine take away to give me this sharpness?
Is the wood grain still wood, or is it a dream of wood? Is the skin still skin, or is it a plastic mask? My sneezing fit eventually stopped, leaving me in that quiet, post-storm clarity. I looked back at the screen, at the reconstructed photo of my mother.
I hit ‘undo.’ I went back to the original-the grainy, slightly yellowed, 72dpi scan. It was blurry. It was small. But when I looked at it, I could actually smell the hairspray she used to use. I could see the specific, uneven way she applied her lipstick that day.
The machine couldn’t reconstruct those memories because it didn’t know her. It only knew “faces.” We must be careful not to become a society that only knows “faces.” We need to remain a society that knows people.
The tool is a shortcut, but the shortcut shouldn’t become the destination. We have to keep our eye trained on the grit, the grain, and the glorious, low-resolution mess of being alive.
Because if we don’t, we might wake up one day and realize we’ve upscaled our entire lives into something sharp, bright, and completely unrecognizable.