Why the kitchen keeps the cook
Institutional cooking is bulk production on a deadline. A school, hospital or nursing-home kitchen cooks food in quantity for a fixed service time, then portions and serves it to students, patients or residents. Software can plan the cycle menu. It cannot stand at the kettle, taste the sauce and decide the batch needs ten more minutes.
The second reason is the room itself. Cooks clean and inspect equipment and work areas, check that holding temperatures hold, and fix small problems before service: a steamer that stalls, a delivery that arrived short, a tray line backing up. Those are judgment calls made in seconds, in a wet and crowded space. Our share of task time that still needs a person is 75% of this job’s work.
Diets add another layer. In hospitals and care homes, meals have to match allergy, texture and sodium restrictions for named people. A system can flag the restriction. Someone still has to make the puree, keep it separate and get the right tray to the right room.
What AI does, what it helps with, what stays with people
Where AI already handles the task, the work is paperwork, not cooking: logging temperatures and holding times, and tracking stock levels against what was served. That slice is 0% of this job’s task time. Both tasks are records that a sensor or a spreadsheet can keep better than a clipboard.
The assisted slice is larger in reach than in hours. Forecasting how much to prepare for a given day, and planning menus that meet nutrition standards and a tight budget, both get faster with software that reads past service counts. A cook still signs off. Our assisted share sits at 25% of task time, and our overall coverage figure, explained on how coverage is measured, is 16 out of 100.
The rest sits with people. Cooking in quantity, apportioning and serving, cleaning and inspecting kitchen equipment, and training or directing other kitchen staff all stay hands-on. About 58% of the task mix we score for this job is physical work, and the robotics tier we record is fixed automation: machines bolted in one place doing one step, not a machine that walks the line.
What the evidence actually shows
There is no direct head-to-head test of AI against institutional cooks in the evidence we hold. Our evidence grade for this job is D, and a D grade means quality parity has not been measured, so we publish no parity number for it. The grading scale is set out on how quality parity is graded.
What would settle it is specific: a timed trial in a real production kitchen, comparing a cook and an automated line on the same menu cycle, scored on food safety holds, waste, portion accuracy and how both handle a short delivery or a broken steamer. Until something like that is published and dated, the honest answer is that the question is open. Our full method is at how we score jobs.
When this could change
Most likely after 2044 (8 in 10 of our scenarios). What that window measures is explained on the replacement-year method.
Two things could pull it earlier. Large central kitchens that cook one menu for many sites are the best case for fixed automation, because the layout stays the same and the volume justifies the machine. And the cost gap matters: the tooling range we record for this job is far below the staffing range shown in the costs panel above, so any machine that reliably takes a whole step becomes easy to justify.
Two things hold it back. Most of these kitchens are small, old and laid out differently from each other, so equipment has to be refitted site by site. And pay is modest: median pay for this job was $37,450 in May 2024 (BLS), which stretches the payback period on a six-figure install. Employment sat near 441,050 and BLS projects about 2.8% growth over the decade to 2035, so demand is steady rather than shrinking.
How to stay needed in an institutional kitchen
Lean into the parts of the job that no forecast can do for you. Run production cooking for volume and timing, so service never slips. Own therapeutic and allergy diets, where a mistake has a name attached. Take on equipment checks and cleaning sign-off, because food safety records are only as good as the person doing the walk.
Two skills are worth adding. First, food safety and HACCP documentation, including using the digital logs that are replacing the paper ones. Second, supervising and training: directing a shift, teaching a new hire the tray line, and planning labor around a delivery that did not show.
What to do: ask to be the person who sets up and checks the kitchen’s forecasting or inventory system, rather than the person it reports on.
Nearby work worth comparing: restaurant cooks, chefs and head cooks, and food preparation workers. You can put any two of them next to each other on the job comparison tool, or read across the whole cooks and food preparation family. For the setting rather than the title, see our schools sector page, and for the machinery question, what humanoid robots can do in physical jobs. If you want the wider view of hands-on work, the list of jobs that mostly need a person is the place to start. Our headline figure for this job is 79 out of 100 (higher is safer).