Why the camera still has a person behind it
Most of this work happens in a room with other people, in real time. An operator frames the shot while the action moves, pulls focus as a subject steps forward, and holds the composition a director asked for ten seconds earlier. Automated systems can track a face reliably. Deciding that the better shot is the reaction rather than the speech is a judgment call made in the moment, and that is the part that keeps a person on the camera.
Then there is the gear. Camera rigs get built, balanced, mounted and struck in cramped studios, on location, on vehicles and in weather. Heads get leveled, cables get run, matte boxes and filters get swapped between setups. None of that is a single repeatable motion, which is why the robotics read on this page sits in the dexterous humanoid tier rather than the fixed-arm tier.
The money side cuts both ways. Camera operators earn a median of $74,990 a year and about 21,550 people hold the job in the US, with employment projected to change by 1.3% between 2025 and 2035 (BLS, 2025). That is a small, skilled workforce rather than a large back office, so there is less of the volume that usually pulls heavy automation spending. Robotic studio pedestals and tracking heads already exist in news and sports, and they have reshaped crew sizes there without clearing the floor.
What software handles, what it assists, and what stays on the operator
Start with what tools can take the lead on. Automated tracking and framing in a fixed studio setup is the clearest case: a preset shot, a known subject, a locked lighting plan. Auto-exposure and auto-focus systems now hold a usable image through moves that once needed a focus puller’s full attention. Tasks grouped this way account for 0% of the time our method assigns to automatable work on this job.
Assisted work is the bigger story on set. Motion-control heads repeat a move exactly on take after take, but a person designs the move. Software flags soft focus, clipped highlights or a dropped frame while recording, and logging tools sort and label the day’s footage before anyone opens an edit timeline. That assisted group covers 27% of automatable task time, which is the honest shape of change here: fewer small chores, not fewer sets.
What stays with people is the rest, and it is the majority: 73% of total task time. Conferring with the director and the lighting crew about coverage, choosing lens, angle and movement for a scene, operating handheld through a live event, and troubleshooting a rig that fails five minutes before a take all sit here. Across all tasks, the share of time AI can handle today comes out at 20; the coverage method explains how that is counted.
What to do: learn the automated systems on your own kit well enough to set them up and override them, because the operator who can do both is the one who keeps the call sheet.
What has actually been tested
There is no published test that puts AI against camera operators on this job’s real tasks. That is why the evidence grade for quality parity here is D, and why we publish no parity number for this occupation. Generated video clips and automated framing demos are not the same as a graded comparison on framing, focus and coverage under production conditions.
What would settle it is specific: a blind comparison where a tracking or generated system and a working operator shoot the same scripted scene and the same live event, scored by directors and editors on usable takes, continuity and coverage. Until something like that is published, treat claims in either direction with care. The quality parity method sets out what counts as a test, and the full scoring method covers the rest.
When the picture could change
Most likely after 2038 (8 in 10 of our scenarios). Two things could pull that earlier. One is cheap, reliable robotic heads spreading from news studios into corporate, streaming and event work, where budgets are tight and setups repeat. The other is generated video getting good enough that some clients skip the shoot entirely for simple explainer and product content, which removes jobs rather than automating them.
Two things hold it back. Hardware that can rig, balance and operate on an unpredictable set is still expensive and slow to deploy, and the physical share of this job keeps it in the harder robotics tier. Liability and craft standards also matter: on a union set, a studio production or a live broadcast, someone has to answer for the shot. The replacement-year method explains what the window is measuring.
How camera operators stay needed
Lean into the tasks that hold the most human time on this page. First, shot design with the director: showing up with a lens and movement plan for a scene, not just executing one. Second, live operating, where the action does not stop and recovery from a missed cue has to happen on the fly. Third, on-set problem solving, from a failing rig to a location that will not light the way the plan assumed.
Two skills compound on top of that. Lighting literacy, because exposure decisions drive the look more than the camera body does. And multi-camera and remote systems work, including setting up tracking and motion-control heads and knowing when to switch them off. Both move you toward the supervisory end of the crew, which is where fewer entry-level slots are being filled.
Adjacent work is worth checking too. The nearest jobs by task are film and video editors, photographers and audio and video technicians. You can put any two side by side on the compare tool, see the wider group on the media and communication equipment workers family page, or look at the whole arts and entertainment sector. If you are weighing a move, the jobs that mostly need a person list is a useful next stop.