Why this job stays in the room
An orderly moves people, not files. Lifting a patient from a bed onto a gurney, steadying someone who is dizzy, easing a wheelchair through a crowded corridor: each step depends on reading weight, pain and balance in the moment. Software can book the trip and pick the route. It cannot take the weight or notice that a patient has gone quiet and gray.
The rest of the shift is just as physical. Cleaning and disinfecting rooms and equipment, changing linens, restocking supply closets, carrying specimens and gear between floors. Autonomous carts already roll through some hospital corridors with linens and supplies, but they follow mapped routes and stop when something unexpected sits in the way. A stripped isolation room still needs hands, cloths and a person who knows what gets wiped twice.
Scale matters here too. The Bureau of Labor Statistics counts about 52,440 orderlies in the United States, with median pay of $38,290 and projected employment growth of 4% from 2025 to 2035 (BLS, 2025). That is a steady, in-person job in hospitals and nursing homes, not a desk role being quietly rewritten by chat tools.
What software handles, what it assists, and what people keep
Start with the part machines can run end to end. On this job that share is 0%. It covers the clerical edge of transport: logging a request, closing out a completed move in the system, keeping a queue in order. None of it touches the patient.
Next comes the assisted group, at 4% of task time. This is where dispatch tools pick which transport goes first, where supply tracking tells you what is missing before you walk the floor, and where a robot cart carries the heavy load of linens while the orderly does the clinical part of the errand. The work still happens, with fewer wasted trips.
Everything else sits with a person: 96% of task time. Transfers and repositioning, helping with bathing and toileting, calming a frightened or confused patient, terminal cleaning of a room after discharge. The task mix is why the task-coverage figure on this page, 2 out of 100, sits where it does; the Can AI Do It? method explains how that share is built.
What has actually been tested
Nothing has yet tested AI or robots against orderlies head to head on their own work. That is why our quality grade for this job is D, the grade we use when no direct comparison exists. We give no parity number for untested work, and you should treat any score elsewhere that claims one with care. The Is It Better Than A Person? method sets out what each grade means.
What would settle it is specific: a hospital trial that puts robotic transport and handling systems against trained orderlies on the same wards, measuring patient-handling injuries, time per transport, infection-control outcomes and what patients say about the experience. Supply-delivery robots have been deployed in hospitals for years, but deployment is not a comparison. Until someone runs that study and publishes it, the honest answer is that the evidence is thin and the physical case is strong.
When the picture could change
Most likely after 2045 (8 in 10 of our scenarios). The When Could It Be Replaced? method explains how that window is produced and what the spread means.
Two things could pull the date earlier. First, hardware: this job’s robotics tier is dexterous humanoid, and any serious jump in cheap, reliable two-armed machines that can lift a real human body safely would change the math. Second, staffing pressure. Hospitals already run short-handed, and tools that cut walking time get adopted fast when the alternative is unfilled shifts.
Two things hold it back. Cost is the first: the hardware and support needed to do the physical part of this job is not priced against a cheap alternative, and the comparison table on this page shows why that gap matters. The second is risk. Dropping a patient is a different category of failure from a bad sentence in a report, so safety rules, infection control and liability all slow any handover. For more on where machines stand on physical work, see our guide to humanoid robots and physical jobs.
How to stay needed as an orderly
Lean into the parts of the shift that stay with people. Safe patient handling and transfers are the core: be the person who can move a heavy, frightened or post-surgical patient without hurting either of you. Patient comfort and de-escalation come next; a confused patient who trusts you is a shorter, safer transport. Third, infection control cleaning, where doing it right and documenting it is a clinical task, not a chore.
Two skills pay off. One is certification in mechanical lift and safe-handling equipment, because that is where human judgment meets the machine. The other is clinical observation and clear handoff: noticing a change in skin color, breathing or alertness and reporting it in the words the nurse needs.
What to do: ask your employer which transport, dispatch or delivery systems they are piloting, and volunteer to be trained on them first.
If you want to move sideways or up, the closest work is nursing assistants, psychiatric aides and home health aides. All three sit in the same aides and assistants family, and all three keep the hands-on core. You can put any two of them side by side with the job comparison tool, look at how hands-on roles rank on our list of jobs AI handles least, or read how the scoring works before you trust any of it.