Being right about quality won't save you: a photographer's warning about the AI backlash

I need to confess something, because it might save somebody a few years of their life.

Back when I was a portrait photographer, I was a gatekeeper. A finger-wagger. One of the people who stood at the edge of a profession and quietly decided who was allowed in.

I didn't think of it that way at the time. Nobody does. I thought of myself as someone who had earned something, and I had. I had spent years learning light, composition, posing, how to make a squirming toddler laugh at the exact right half-second. I had put about $30,000 of my own money into a Nikon D3, a bag of lenses, and professional lighting. I had put in the reps. So when someone held up a phone snapshot and called it a "photo," part of me bristled. That's not photography. That's a picture. There's a difference, and I'm one of the people who knows it.

I said the thing all of us said. A phone will never replace a real camera. I was right about the technology and badly wrong about what mattered. The distance between those two things is the most expensive lesson I have ever learned, and I want to give it to you for free.

Being right is not the same as being safe

Here's what is strange, looking back: the gatekeepers weren't crazy. We had evidence. For years, the phone genuinely couldn't do what a pro could.

You can still find our old certainty fossilized all over the internet, and it reads like a time capsule of people about to be very surprised. "Their camera will never replace a DSLR. It's just a good phone camera, nothing more," one commenter wrote under a professional photographer's iPhone review, in 2017. A working photographer wrote, a couple of years ago, that he could say "without a shadow of a doubt: smartphones will never replace cameras." Reviewers reassured each other that phones were "still inferior in comparison to what dedicated, traditional cameras can do." Even a phone maker hedged it perfectly back in 2010: a phone would "never replace the specialist camera, but for most people, a good snap is good enough."

Read that last one again, because we said it as a defense and it was the whole obituary: for most people, a good snap is good enough.

That was the trap. We kept arguing about the ceiling, what the best camera in the best hands could do, while the market was being decided at the floor. The customer was never grading my work against a phone on technical merit. The customer wanted the moment, and the moment in their pocket was free. While we congratulated ourselves on bokeh and dynamic range, the numbers moved underneath us. Worldwide shipments of dedicated cameras eventually fell about 94 percent from their 2010 peak. Kodak went bankrupt the same year a billion dollars changed hands for Instagram, a two-year-old photo app. More than 2,000 Sears and Walmart portrait studios closed almost overnight in 2013, and more than 4,300 people lost their jobs.

The gatekeepers were right about quality the entire way down. Being right about quality didn't save a single job. Mine included.

Now I work in tech, and I watch new gatekeepers do my old job, in at least three different rooms at once. Let me walk you through all three, because if you work for a living in the 2020s, you are standing in one of them.

Room one: the engineers and "vibe coding"

The dirty word in software right now is vibe coding: building things by describing what you want to an AI in plain English and letting it write the code. The reaction from a lot of the developers I work with and genuinely admire is visceral. The nose goes up. The term gets said with air quotes and a little curl of the lip. That's not engineering. That's not real coding. There's a difference, and I'm one of the people who knows it.

I know that bristle. I have felt that exact bristle. I felt it holding a Nikon.

Here's what makes it so seductive and so dangerous: the gatekeepers aren't wrong on the facts. One analysis of 153 million changed lines of code projected that "churn," code that gets written and then thrown away or rewritten within two weeks, would roughly double as AI assistants took over. A separate study of around 800 developers found that teams using an AI coding assistant shipped about 41 percent more bugs. The U.S. cybersecurity agency has published guidance warning that building software with these tools brings new security risks and widens the ways a system can be attacked. Senior engineers even have a name for what they fear they are becoming: "code janitors," cleaning up after people who shipped a thousand lines they never read.

Those are real concerns from real experts who paid real dues. I want to honor that completely. If you spent ten years learning to architect systems and debug a cryptic failure in a legacy codebase at 2 a.m., that is hard-won knowledge, and a chatbot producing plausible-looking code does not erase it. You earned that the way I earned my eye for light.

But none of that protected me. Being right about the craft was the comfort that kept me from seeing the change until it had already happened. "Is the AI as good as a real engineer?" is the bokeh-and-dynamic-range argument all over again. The question that decides your future is the quieter one: for most of what most people need built, is good-enough-and-instant going to beat excellent-and-expensive? I have seen that movie. I know how it ends.

Room two: the product people, and yes, that includes me

Let me stand in my own room, because I don't want anyone thinking I'm pointing only at the engineers. I work as a product owner, the person who decides what gets built and why. And product owners are gatekeeping just as hard. We just don't have a catchy insult to rally around.

That is almost a sleight of hand. "Vibe coding" is a meme. It is mockable. It gave developers a flag to plant and a word to sneer. What is happening to my kind of work has no nickname, so it is happening quietly, which makes it more dangerous, not less. Nobody wrinkles their nose at vibe coding's product-side cousins, because those cousins don't have names that land. Try saying them out loud:

• Prompt-written specs: describing a feature to an AI and letting it draft the requirements document, instead of laboring over it for two days.
• Automated requirement-gathering: pointing a tool at the recordings and transcripts of your planning calls and letting it pull out the real asks, the conflicts, the edge cases.
• AI roadmapping: feeding a model your backlog, your support tickets, and your priorities, and letting it propose the order of work.

None of those have the ring of vibe coding. You can't say "AI roadmapping" with a curl of the lip and get a laugh. So the gatekeeping in my world shows up as something subtler and more flattering: a real product owner develops intuition. A real one sits in the room and reads the politics. You can't automate trust. You can't prompt your way to product sense.

Say it with me: that's not wrong. Product judgment is real. Reading a room is real. The relationships are real. I have all of those and I am proud of them.

It is also exactly what I said about my photography eye. It is the precise shape of the comfort that sank me. The people who pay a product owner don't ultimately buy your intuition. They buy clear decisions, made well, made fast. The moment someone with AI tooling can turn a messy hour of planning calls into a clean, ordered plan in ten minutes while I'm still scheduling the follow-up meeting, my hand-crafted version starts to look like a $30,000 lighting kit in a world of phones: lovely, expensive, and increasingly beside the point to the person paying for it.

The engineers at least got a villain with a name to argue about. The product people got silence, which is much easier to lose to.

Room three: the media, the oldest gatekeepers of all

If you want to see where the first two rooms are heading, look at the room that has been on fire the longest: the press.

For a century, "journalist" was a gate with a guard on it. You needed the byline, the credential, the institution, the printing press or the broadcast license. The gatekeepers decided what was news and who got to deliver it. When social media handed every person a printing press in their pocket, the institutions reacted the way I reacted to the iPhone: that's not journalism. That's just people with opinions. There's a difference, and we're the ones who know it.

Then, in November 2024, Elon Musk posted four words that ran straight through the whole argument: you are the media now. He told his followers to post what they saw, correct each other, and route around the legacy outlets. Whatever you think of the messenger, and there is plenty to scrutinize, including one analysis that flagged scores of his own election posts as false or misleading and viewed billions of times, the phrase landed because the ground had already shifted under it.

Look at the numbers the gatekeepers were standing on. In 2025, Gallup found trust in mass media fell to 28 percent, below 30 percent for the first time in the poll's roughly fifty-year history, down from the 68 to 72 percent it held in the 1970s. The share of Americans with no trust at all in the media now exceeds the share who trust it a fair amount or more. Meanwhile a single podcast episode, Donald Trump on Joe Rogan's show, drew more than 26 million views in its first day, a number most cable news shows can only dream about. The audience didn't just drift away from the institutions. It walked into the arms of "unqualified" amateurs with microphones and phones.

Here's the gatekeeper moment I can't stop thinking about, because it is me in 2010, word for word. The CEO of Axios, a real and serious journalist, took the stage at the National Press Club and said of Musk's slogan: "You having a blue check mark, a Twitter handle, and 300 words of cleverness doesn't make you a reporter." Musk replied: "Yeah, whatever. You are the media now, and the legacy media know it."

The Axios CEO is right. A blue check isn't a press pass. The hard work of real reporting is real, and a world without it is poorer and more dangerous. I am not cheering the bonfire. But I have heard that exact tone of voice before. I have used it. "A phone doesn't make you a photographer." "A prompt doesn't make you an engineer." "A Twitter handle doesn't make you a reporter." Every one of those sentences is true, and not one of them has ever stopped a market from deciding that good-enough-and-free beats excellent-and-expensive. I planted my flag on that exact hill and watched the tide come in over it.

What all three rooms get wrong

Here's the pattern, once you have been burned badly enough to see it.

The gatekeeper's mistake is never being wrong about the craft. The photographer was right that phones shot worse photos. The engineer is right that vibe-coded apps break. The product veteran is right that intuition matters. The journalist is right that a tweet isn't reporting. They are all correct, and their correctness is the trap, because it lets you fight the battle on the one ground where you are winning (quality, rigor, dues paid) while you lose the war on the ground that decides things: what the customer, the business, or the audience will happily accept for free or cheap.

You can win every argument about the ceiling and still get buried by the floor.

To the gatekeepers: look up before the tide does it for you

If you recognize yourself in any of those three rooms, I'm not telling you your skills don't matter. They matter enormously, maybe more than ever. I'm telling you what I wish someone had grabbed me by the collar and said in 2010.

Don't let "I'm right about the quality" become the pillow you fall asleep on. That was the most comfortable and most expensive belief of my life. The photographers who survived the smartphone weren't the ones who won the quality argument. They were the ones who stopped having it, picked up the new tools, met people where they were, and aimed their hard-won skill at the work the phone couldn't copy.

Here's the encouraging part: in this change, your experience is an advantage in using the new tool, not just a thing the tool threatens. The traits that make AI coding tools powerful, specifying a problem precisely, reviewing the output critically, knowing what good looks like, are the traits of senior engineers already. The same holds for product: the person who can tell when the AI's plan is subtly wrong is the person who spent a decade building plans. The same holds for journalism: the reporter who knows how to verify, source, and add context is exactly who should be using these tools against a sea of slop. AI tends to amplify whatever judgment you already have. The person who learns to direct it becomes far more capable, not less. The danger is reserved almost entirely for the gatekeeper who refuses to touch the thing on principle, who, like me, defends the darkroom so hard they are the last one standing in it.

I hung up my hat to protect what I loved, and I don't fully regret it. But make no mistake: I left. I did not adapt. I'd rather you adapt.

To everyone being sneered at: stop apologizing for building

Now let me talk to the other people in these rooms, the ones on the receiving end of the nose-wrinkle. The self-taught builder shipping an app by describing it. The junior who lets AI draft the first version of a spec so they can spend their time on the actual humans. The citizen with a phone and a newsletter covering their own city council better than anyone has in years. The people being told, in a hundred small condescending ways, that what they are doing is naughty. Cheating. That they have not earned the right.

I want to say this as clearly as I can, because I was once the voice making people feel exactly this way, and I'm ashamed of it: do not let the gatekeepers hand you imposter syndrome.

The pattern is so old it's almost funny. Stack Overflow, the temple of professional programming, wrote that for experienced developers, vibe coding "holds the existential threat of imposter syndrome." Sit with that. The anxiety isn't only that you might be a fraud. It's that they might be replaceable, and shame is the oldest tool ever invented for protecting a moat. When someone makes you feel small for building something that works, check whose insecurity is doing the talking.

Here's the truth the gatekeepers won't say out loud: you made something where before there was nothing. The designer who never set up a build pipeline shipped a working app. The founder who couldn't afford a developer has a prototype. The kid with an idea and no diploma has users. That isn't nothing. That's the entire game. Remember: the number of photography businesses in this country went up after the smartphone, because a flood of supposedly unqualified people discovered they could make real work and real money. The gatekeepers swore that would degrade everything. Instead it opened the craft to everyone.

You are not naughty for using the best tool available to you. You are early. There's a difference, and now I'm one of the people who knows it.

One honest word in the other direction, so this isn't just a pep talk: the way to disarm a gatekeeper isn't to dismiss the craft they are protecting. It's to out-learn their lowest expectation of you. Use the AI as a teacher, not just a vending machine. Ask it why the code works, why the plan is ordered that way, how to check a claim before you post it. The builders who interrogate the output instead of just shipping it are the ones still standing when the easy magic wears off. Don't be the cautionary tale the gatekeepers point at. Be the reason their argument stops working.

The bridge nobody's standing on

Here's what gets me about all three rooms: the gatekeepers and the newcomers are each half right, and they are going to miss the same enormous opening by fighting each other instead of looking at the open field between them.

The veterans are right that judgment, rigor, and accountability matter. The newcomers are right that access, speed, and "I made a thing that works" matter. The people who win this era, the way the app developers and the small-business owners and the thriving phone-era freelancers won the last one, won't be the purists or the people who think craft is obsolete. They will be the ones standing on the bridge between: enough fluency with the new tools to move fast, enough hard-won judgment to know when the fast thing is about to blow up at 2 a.m. or mislead a million readers.

That bridge is wide open right now, precisely because everyone is too busy sneering across it.

I spent the smartphone revolution on the wrong side of that bridge, arms crossed, certain that being right about quality made me safe. It didn't. The technology I dismissed went on to create whole categories of work and wealth I couldn't have imagined from where I stood, and I watched all of it from the sidelines, because I had decided the new way was beneath the craft I loved.

I don't get that decade back. But you, whichever room you're in, whichever side of it you're on, you still get to choose.

To the gatekeepers: the moat you're defending is real, but the castle is already being rebuilt somewhere else. Come help build it.

To the newcomers: build loudly, learn hungrily, and let them wag their fingers. Finger-wagging is just the sound a moat makes while it's draining.

I should know. I made that sound for years.

Sources

• Dedicated camera shipments down about 94 percent from their 2010 peak, from CIPA data (PetaPixel)
• Kodak's January 2012 bankruptcy (ABC News)
• Facebook's 1 billion dollar purchase of Instagram, April 2012 (Meta)
• CPI Corp closing more than 2,000 Sears and Walmart portrait studios, with more than 4,300 jobs lost, in 2013 (NPR)
• An analysis of 153 million changed lines of code projecting that code "churn" would roughly double (GitClear)
• A study of around 800 developers finding teams using an AI coding assistant shipped about 41 percent more bugs (Uplevel, via DevOps.com)
• U.S. cybersecurity guidance on the security risks of building with AI (CISA)
• Stack Overflow on vibe coding and "the existential threat of imposter syndrome" (Stack Overflow)
• A text-to-app startup (Lovable) reaching a billion-dollar valuation about eight months after launch (TechCrunch)
• "Vibe coding," 2025 Word of the Year (Collins Dictionary)
• Elon Musk's "you are the media now" post, November 2024 (Elon Musk on X)
• An analysis flagging scores of Musk's election posts as false or misleading, viewed billions of times (TechCrunch, reporting the Center for Countering Digital Hate findings)
• Trust in mass media at 28 percent, below 30 percent for the first time, down from 68 to 72 percent in the 1970s (Gallup)
• Trump's Joe Rogan episode drawing more than 26 million views in its first day (Newsweek)
• The Axios CEO's National Press Club remark, and Musk's reply (TheWrap)
• The number of U.S. photography businesses rising in the smartphone era (IBISWorld)