■ MODERATE RISK ■ Creative Arts
Half the market, yes. Generated imagery has already gutted stock and much of commercial product work, while anything requiring you to physically be somewhere real remains firmly a photographer's job.
“AI generates images. But capturing a real moment — a wedding, a war zone — requires being there.”
Our AI replacement risk score — how we score jobs
Commercial photography has always had two sources of value: the technical craft of making an image, and being present at something that actually happened. Generative models attacked the first with startling speed. Stock libraries — the reliable passive income tier of the industry — now compete with unlimited on-demand imagery at effectively zero marginal cost. Product photography for e-commerce, catalogue shots, generic lifestyle imagery for brochures and websites, and concept work for pitches are all being generated or heavily augmented, and clients who once budgeted a shoot day now budget a prompt.
Editing has changed too. Generative fill extends backgrounds and removes distractions, AI culling picks the sharp frames from three thousand exposures, automated retouching handles skin and colour work that once billed hours, and denoise tools have quietly extended the usable range of every camera. For a working photographer this is mostly good — it compresses post-production, historically the unpaid half of the job.
What holds is documentary reality. A wedding happened once, and the couple wants photographs of their actual grandmother. Photojournalism depends on the image being evidence, and that value increases as synthetic imagery makes provenance scarce — which is why authentication standards are being built into cameras. Portraiture, event, sports, and real-estate work all require someone with a camera in a place. Add the fact that clients hire photographers for direction and reassurance as much as for shutter presses. Our risk score of 45 captures an industry losing its commodity half and keeping its evidentiary one.
Automatability: our editorial assessment of current and near-term AI capability
Already biting hard, with the worst falling on stock and commodity commercial work since the mid-2020s. Through 2030 expect e-commerce product imagery to be predominantly synthetic or hybrid, and generic marketing imagery to stop being commissioned at all. Event, wedding, documentary, and press photography remain human indefinitely, with provenance and authentication becoming a selling point. The industry ends up smaller, more polarised, and less friendly to generalists.
Largely, as a business for individual contributors. Buyers who need a generic image of a smiling team in an office can generate one instantly, exclusively, and free of licensing questions. Stock still has a market for authentic documentary and editorial imagery where provenance matters, but the passive-income model built on volume uploads of generic frames is effectively finished.
Among the least at risk in the field. The product is documentation of a real event with real people, which cannot be generated after the fact. AI helps these photographers rather than threatening them, by cutting culling and retouching time dramatically. The competitive pressure in that market comes from oversupply of photographers, not from software.
Yes for editing efficiency — culling, retouching, background cleanup — and increasingly with disclosure for compositing. No for anything sold as documentary or journalistic, where fabricating elements destroys the entire value proposition and, in press work, ends careers. Agree with clients up front what level of manipulation is acceptable for the deliverable.
Anything evidentiary or relational: weddings, events, sports, photojournalism, portraiture, architecture and real estate, and industrial documentation. Also high-end commercial work where a specific creative vision is being bought. Avoid building a business on generic stock, catalogue product shots, or interchangeable lifestyle imagery — those markets are already being served by generation.