AI vs Photographers: What Happens to Photography When AI Can Make Anything?
Recently, I typed "Peruvian woman, tall, blonde, white, late twenties, photographer, Gold Coast beach, golden hour, Fuji film look" into the Meta AI image generator, mostly to see what it thought of me.
It gave me someone with better skin and a way nicer hair, standing on a beach that doesn't exist, holding a camera with the strap going into her neck. It was, in a flat sort of way, lovely. It was also nobody, nowhere, at no particular time, and I found I had nothing to say about it.
That's roughly my whole position on AI and photography, but it deserves more than an anecdote, so here's the longer version. What's changing, what it means for commercial, fashion, product and brand photography, where AI helps me, and where I think human photographers are about to become more valuable rather than less.
First, the honest concession
AI image generation is very good and getting better by the minute. For a lot of what used to be paid photography work — generic illustrative images, concept art, backgrounds, mock-ups, "a laptop on a desk with a plant" — it is faster, cheaper and often perfectly adequate. The stock photography business is being eaten from the bottom, and some commercial photographers who made a living shooting generic lifestyle imagery for licensing are genuinely losing work. Pretending otherwise would be dishonest and, as a bonus, not very useful to you.
It's also useful to photographers. I use AI tools most weeks: masking tools in Lightroom that would've taken twenty minutes by hand, generative fill to remove a fire exit sign from a hotel lobby wall, transcription for client calls, first drafts of shot lists. It's made the boring 30% of the job faster. Nobody's mourning the fire exit sign.
So the question isn't "AI or photographers". It's: which parts of photography were ever about the image, and which were about something else?
Generating an image vs documenting reality
This is the distinction everything else hangs on.
A generated image is a plausible picture. A photograph is evidence that something was in front of a lens at a moment in time. They can look identical and they are different kinds of object, the way a painting of a receipt and a receipt are different, even if the painting is better.
For a lot of imagery, plausibility is all anyone wanted. For a lot of commercial imagery — the kind businesses use to make people trust them — evidence was the entire point, and people just didn't have to think about it until now.
Which is why the reaction to AI in advertising hasn't been "that looks fake". It's been "you lied to me".
What happened when brands tried it
The pattern so far is remarkably consistent.
When Guess ran an AI-generated "model" in a 2025 Vogue advertisement, the backlash covered beauty standards, the models who weren't hired, and the general sense that a fashion brand had quietly swapped a person for a rendering. Mango drew the same response to an AI campaign in 2024, and H&M's 2025 move to create "digital twins" of real models raised the questions of consent and compensation that are going to define the next few years of this. Levi's had to clarify its position on AI models back in 2023 after announcing it would use them to "increase diversity" — a sentence that didn't survive contact with the public.
Notice what these have in common: fashion and luxury, where the customer is paying precisely for craft, taste, and the implied presence of humans who care. The more a brand sells human judgement, the more expensive it is to be caught faking it.
The legal side is catching up too. Getty Images sued Stability AI over training data; the US Copyright Office has repeatedly held that purely AI-generated images can't be copyrighted, which means a brand "owns" its AI hero image in a much weaker sense than it owns a commissioned photograph. And Adobe, Leica, Nikon and others are now shipping Content Credentials — cryptographic provenance data that lets a viewer check whether an image was captured or generated. Within a few years, "this is a real photograph" is likely to be a verifiable label, not a vibe. That changes what real photographs are worth.
The strange new value of imperfect content
Here's the development I didn't expect.
For fifteen years the direction of travel in commercial imagery was more polished. Better retouching, cleaner sets, more flattering light. Then generation arrived and could produce infinite polish for free, and polish stopped meaning anything. A flawless image is now the cheapest kind of image to make.
What's scarce is the opposite. The slightly grainy frame. The hotel room with the real, not-quite-centred lamp. The founder with the crooked smile she's had since school. The tradie with concrete on his boots. These read as true because a generator wouldn't have chosen them, and "true" is suddenly the premium attribute.
This is the bit I'd underline for any business owner reading: the imperfect content you've been embarrassed to post — the phone clip, the behind-the-scenes, the photo where the light's a bit off but it's really your shop — has quietly become more valuable than the polished stuff. Done and honest beats perfect and invisible, and now it beats perfect and synthetic too.
Where human photographers keep their value, sector by sector
Fashion. Campaign imagery of generic models is under real pressure. But a fashion lookbook photographer is documenting the actual garment — how the linen falls on a real body in real light, which is what the buyer at the boutique and the customer at home both need to know. Behind-the-scenes content for fashion brands has moved from nice-to-have to core, (I offer these sessions and call them Documentary Sessions") because it's the part that proves there's a studio, a pattern-maker, a fitting. The more synthetic the industry's surface gets, the more a brand gains from showing the hands that made the thing.
Hotels and retreats. A generated image of a hotel room is, functionally, a lie to a guest who will stand in that room. A hotel brand photographer's value was always partly evidential — this is the view, this is the pool, this is the breakfast — and that's now the whole value. Retreat photography goes further: it's documenting something that happened to real people at a real place over real days. You cannot generate the look on someone's face on day three. That's what a retreat photography package is buying, and it's unfakeable.
Founders and personal brands. This one's simple. Your personal brand is your face, your voice, your way of standing in a room. AI can make a face. It can't make yours doing the thing you do, and your customers are buying you. The first time a founder gets caught using a generated headshot, they'll learn what the fashion houses learned.
Trades and construction. Possibly the sector with the most to gain. A construction company photographer is producing proof of work: this site, this crew, this finish, this date. In an industry where the customer's fear is "will they turn up and will it be done properly", generated imagery is worse than useless — it's exactly what a dodgy operator would use. Real progress photos, real handover shots, real faces on real sites are about to be a competitive advantage, and they're cheap relative to the jobs they win.
Product. Split decision. Pack shots on white are being automated and that's fine. But product photography that shows scale, texture, how it sits in a hand, how the label actually looks under kitchen light — that's documentation, and it's what stops returns. A fotógrafo de producto who understands this will be shooting the real object for the same reason a jeweller shows you the actual ring.
Creative businesses. Designers, makers, studios. Your work is the product and your process is the marketing. Documentation of process — the sketches, the kiln, the third prototype — is something AI can imitate in style and never in fact.
Fine art. This deserves its own essay, but the short version: generation has pushed fine art photography back towards what made it interesting in the first place — the decision to point a camera at this and not that, the physical act of being there. Provenance, print, edition and the artist's presence matter more, not less. The gallery market was never buying pixels.
The parts of the job that were never about the camera
When a client books me for a brand shoot, the images are the deliverable but they're not what's being paid for. What's being paid for is:
Direction. Getting a nervous person from "I hate this" to looking like themselves in twenty minutes. I wrote a whole piece on that (what actually happens in the first twenty minutes) and none of it involves the camera.
Judgement about use. Knowing that the website needs a landscape with space on the left and the Reel needs vertical, and that the launch in March needs a frame with a prop in it. Knowing where the content goes. This is why I build websites too; the photography and the site are one decision, not two.
Relationship. Being the person who turns up to the same café every six months and knows the owner's daughter started uni. Photographers who work with a business over years become part of how that business sees itself.
Consent and representation. A real photograph involves real people who agreed to be in it, were paid or credited, and look like the community the business serves. That was always a quiet virtue; now it's a legible one.
Being there. The retreat, the handover, the opening night, the morning the first batch came out of the oven. The photographer is a witness. You can't generate a witness.
How photographers should adapt (a list, learned the hard way)
For the photographers reading, because I know you are:
Stop competing on polish. You'll lose. Compete on presence, direction and trust.
Use the tools for the boring parts. Culling, masking, cleanup, admin. Bank the hours; spend them on clients.
Shoot the evidence. Real locations, real staff, real product, time-stamped, with Content Credentials on if your camera supports them.
Learn where the images go. Websites, ads, email, print. The photographer who understands the content plan is hard to replace; the one who hands over a Dropbox link isn't.
Show your own process. Behind-the-scenes of you is proof that you exist, which is now a selling point.
Put a human in every frame you can. A face, a hand, a shadow. The thing generators get subtly wrong is the thing buyers are learning to look for.
Be local and say so. Currumbin, Burleigh, Tugun. Generators don't have a suburb.
Don't lie about it. If you used generative fill on a background, be prepared to say so. The trust is the product now.
AI vs Photographers: Where this lands
AI will absorb the photography that was only ever about producing a plausible picture. It was always going to — that work was halfway to stock already.
What's left is everything that was about a real person, in a real place, at a real moment, directed by someone who knew what the image was for. That's not the leftover. That's the part that mattered, and for the first time in a while the market is pricing it accordingly.
For a business owner, the practical conclusion is almost boringly simple. Show the real thing. Your face, your space, your crew, your product in actual light. Post the imperfect version. Hire someone to help you do it well, once, in a way that lasts — and then don't apologise for it.
The next step
If you've been wondering whether it's worth getting real photographs now that you could generate something for free: yes, and more than it was two years ago. Send me where you are and what you'd like people to believe about your business when they see it, and I'll tell you what a shoot that documents it — rather than decorates it — would look like.
FAQ
Will AI replace photographers? It will replace a lot of generic, illustrative imagery — stock-style lifestyle shots, backgrounds, concept images. It won't replace photography whose value is evidential or relational: real people, real places, real products, directed by someone who understands how the images will be used.
Can businesses use AI-generated images for marketing? Legally, in most places, yes, though purely AI-generated images generally can't be copyrighted. Commercially it's riskier than it looks: several fashion brands have faced public backlash, and customers who identify AI imagery tend to distrust the claims around it.
Is it legal to use AI images commercially? Generally yes, subject to the tool's licence, but ownership is weaker than for a commissioned photograph, training-data lawsuits are ongoing, and some jurisdictions require disclosure in advertising. Check the terms and, for anything involving a real person's likeness, get consent.
How is AI useful to photographers? Culling, masking, retouching cleanup, noise reduction, transcription and admin. It speeds up the parts of the job that happen after the shoot, which leaves more time for directing and client work.
What's the difference between AI-generated images and real photography for a brand? A photograph documents something that existed in front of a camera; a generated image is a plausible picture of nothing in particular. For brand trust — About pages, team photos, product, venues — the difference is the whole point.
What are Content Credentials? An open standard (from the Content Authenticity Initiative / C2PA) that attaches verifiable provenance data to an image, showing whether it was captured by a camera, edited, or generated. Adobe, Leica, Nikon and others support it.
Comments