People occasionally ask, half joking, when I'm going to stop carrying a full-frame body now that AI-generated content makes up a real chunk of my delivered work. The honest answer is never, and it's worth explaining why, because I think the framing of AI replacing real cameras misunderstands what each tool is actually for.
AI tools are for scale, cameras are for trust
When a brand wants fifty variations of a product shot for ad testing across markets, generating those with AI is faster and cheaper than fifty physical setups, and clients increasingly expect that option. But when a client needs proof that a real event happened, a real person said something, a real product exists in a real hand, a photograph carries a kind of evidentiary weight that generated imagery doesn't have yet and, honestly, shouldn't have. Those are different jobs, not competing versions of the same job.
The gear that hasn't changed
My main body is still a full-frame mirrorless camera I bought specifically for how it handles focus tracking on moving subjects, motorsport mostly, and low-light color rendering that AI upscaling still can't fully fake convincingly under close inspection. A 24-70mm and an 85mm cover the bulk of my paid event work. None of that changed when AI video entered my pipeline. It just sits alongside it now instead of being the whole story.
- Real cameras: anything requiring proof, authenticity, or client trust
- AI generation: scale, variation, and concepting before a real shoot
- The two increasingly feed each other rather than compete
Where the two actually meet
The most interesting workflow I've landed on this year uses real photography as reference input for AI generation, feeding actual shot product photos into a model to keep generated variations consistent with the real object rather than hallucinating details. That's not photography being replaced. That's photography becoming the anchor that keeps AI output honest. I expect that relationship to get closer over the next year, not more adversarial.