Why hands stay on the animals
Animal care is a body job. A shift means feeding and watering, hosing out kennels and runs, bathing and brushing, walking dogs, and moving animals between pens and exam rooms. None of that happens on a screen. A frightened cat under a bench, a dog that bolts at the gate, a sick rabbit that stops eating: each one needs a person in the room who can read the animal and act in seconds.
The second reason is judgment that is built from touch and smell. Caretakers notice a limp before it shows in a record, a coat that feels wrong, a litter box that looks off. They decide when to call a vet and when to wait. Software can flag a pattern in weight or activity data, but someone still has to confirm it with the live animal and choose what happens next.
The scale is steady, too. The Bureau of Labor Statistics counted about 266,910 animal caretakers in the United States, with median pay of $35,360 a year (BLS, 2025). BLS projects employment growth of 12.1% between 2025 and 2035, driven by pet ownership and boarding, shelter work and research facilities. Our share of task time that still needs a person is 80% of the job.
Software, sensors and the parts people keep
AI already takes some of the desk work around the animals. Intake and boarding records, appointment and feeding schedules, adoption listings, vaccination reminders and routine owner emails can be drafted or filed by software. Our share of task time AI can handle without a person is 0%, and it sits almost entirely in paperwork rather than care.
A larger part of the job is assisted rather than taken. Cameras and wearable sensors track activity, feeding, weight and sleep, and flag an animal that is off its normal pattern. Image tools help match strays to lost-pet reports. Automatic feeders and water systems cover timed rounds. Our assisted share is 20%. In each case the tool narrows where to look; a caretaker still looks.
What stays with people is the contact work: restraining and handling animals safely, cleaning and disinfecting enclosures, grooming, administering medication under instruction, and calming animals that are stressed, aggressive or in pain. These are also the tasks that carry liability. An error here hurts an animal or a person, so facilities keep a named human responsible for them. Our overall measure of how much of the job AI can do today is 15 out of 100, and how we measure coverage explains what goes into it.
What has actually been tested
Not much, and that matters. The quality-parity grade for this job is D. That is our marker for no direct test of AI against a qualified animal caretaker on this job’s real tasks, so we publish no parity number for it. Claims that software matches a caretaker are, for now, untested.
What would settle it is specific: a trial where a sensor-and-alert system is scored against experienced staff on catching illness and injury early in a working shelter or kennel, with false alarms counted; and a robotics trial where machines handle cleaning and feeding rounds in an occupied facility for weeks, measured on animal stress, injuries and hours saved. Research on AI in laboratory animal work is active, but it mostly targets study design and testing methods, not the caretaking shift. Until someone runs the caretaking trials, the honest answer is that the comparison has not been made. You can read how we grade evidence on the quality parity method page, and the wider scoring method covers the rest.
When the picture could shift
Most likely after 2043 (8 in 10 of our scenarios). The replacement-year method explains how that window is built.
Two things could pull it earlier. First, cheap general-purpose robots that can open gates, grip a leash and work in wet, cluttered rooms; our robotics panel puts most of this job’s physical work in the dexterous humanoid tier, which is the hardest and least mature. Second, large boarding and research operators standardizing their facilities around machines, the way warehouses did, so the environment bends to the hardware.
Two things hold it back. Live animals are unpredictable and easily hurt, so every physical step carries welfare and safety risk that owners and inspectors take seriously. And the money does not favor machines at this pay level: a facility replacing hands-on rounds has to buy and maintain hardware, while the labor it displaces is among the lower-paid work in the economy. The cost panel on this page shows how those two sides compare.
Good to know: the pressure in animal care is more likely to show up as fewer admin and front-desk hours per facility than as fewer people on the kennel floor.
How to stay needed
Lean into the tasks that stay in the room. Handling and restraint of difficult animals, spotting health problems early and reporting them clearly, and running cleaning and disease-control routines to a standard an inspector would accept. Those three are the backbone of the job, and they are also what employers struggle to cover on weekends and holidays.
Two skills raise your floor. One is medical: medication administration, basic first aid and the vocabulary to hand a vet a useful report. The other is reading the data tools instead of ignoring them, so when a sensor flags an animal you can confirm, dismiss or escalate the alert with reasons. People who can do both become the person a facility builds the schedule around.
Nearby work is worth a look if you want more specialization. Try animal trainers for behavior work, veterinary assistants and laboratory animal caretakers for clinical and research settings, or animal control workers for field and enforcement roles. The animal care and service workers family and the other services sector show the neighboring jobs together.
Our headline figure for this job is 80 out of 100 (higher is safer); what that score means is set out on its own page. To see where animal care sits against other hands-on work, put two jobs side by side on the job comparison tool or browse the list of jobs that mostly need a person.