The exposure report · Food preparation

Food preparation, and the largest occupation nobody discusses

Seventy-nine per cent of food preparation work cannot be sped up. Baristas alone account for 3.1 million people.

Exposure means published research judged an AI could halve the time a task takes, at the same quality. It is not a prediction about any job. How to read this


16
occupations
323
rated tasks
79%
cannot be sped up, 16 of 22 in the economy
50pts
between the most and least protected job here
Finding one

Baristas: 84 per cent protected, and no middle band at all

Sixteen per cent reachable today, zero in the band needing a tool. That is unusual: almost every occupation in the report has a substantial middle band.

It means the annotators judged that no purpose-built software would halve the rest. There is nothing waiting to be built here.

Finding two

Even the host is half protected

Hosts and Hostesses are the least protected occupation in the group at 50 per cent, and Dishwashers are at 100 per cent.

Sixteen occupations, 323 tasks, so this is the thinnest group in the report after Legal. Treat the figures as indicative.

Every occupation in the group, least protected first

Solid is reachable with a chat window today, pale needs software building first, grey cannot be sped up. Task counts are shown because several of these rest on fewer than twenty.

Hosts and Hostesses, Restaurant, Lounge, and Coffee Shop
Buffet Hostess · Dining Coordinator · General Teller · Greeter
Tasks22 Doing it 325k
First-Line Supervisors of Food Preparation and Serving Workers
Cafeteria Manager · Dietary Supervisor · Dining Services Director · Food and Beverage Director
Tasks26 Doing it 1041k
Chefs and Head Cooks
Banquet Chef · Chef · Cook · Executive Chef (Ex Chef)
Tasks21 Doing it 130k
Cooks, Private Household
Certified Personal Chef (CPC) · Personal Chef · Personal Private Chef · Private Chef
Tasks13 Doing it 0k
Waiters and Waitresses
Banquet Server · Busser · Cocktail Server · Food Runner
Tasks25 Doing it 1804k
Food Servers, Nonrestaurant
Food Server · Food Service Worker · Kitchen Runner · Room Server
Tasks14 Doing it 243k
Bartenders
Banquet Bartender · Bar Captain · Bartender · Mixologist
Tasks20 Doing it 485k
Cooks, Restaurant
Back Line Cook · Banquet Cook · Breakfast Cook · Cook
Tasks20 Doing it 1194k
Cooks, Institution and Cafeteria
Cafeteria Cook · Cook · Dietary Cook · Dinner Cook
Tasks17 Doing it 393k
Baristas
Barista · Catering Barista
Tasks19 Doing it 3095k
Fast Food and Counter Workers
Cafe Server · Cafeteria Server · Cafeteria Worker · Counter Worker
Tasks28 Doing it 3095k
Food Preparation Workers
Cook · Cook Aide · Deli Clerk (Delicatessen Clerk) · Diet Aide
Tasks31 Doing it 783k
Cooks, Fast Food
Cook · Deep Fat Fryer Operator · Fast Food Cook · Fry Cook
Tasks19 Doing it 768k
Cooks, Short Order
Broiler Cook · Cook · Deli Cook (Delicatessen Cook) · Food and Beverage Attendant
Tasks11 Doing it 125k
Dining Room and Cafeteria Attendants and Bartender Helpers
Barback · Buffet Attendant · Bus Boy · Bus Person
Tasks24 Doing it 337k
Dishwashers
Busser · Dish Machine Operator (DMO) · Dish Room Worker · Dish Runner
Tasks13 Doing it 377k
All 16 food preparation occupations combined
Task-weighted across the whole group, so a large occupation counts for more than a small one
Tasks323
Reachable today Needs software built first Cannot be sped up
If you work in People

The clearest example of an empty middle band

Every chapter in this report argues that the middle band is a backlog. This is the one group where it is genuinely close to empty.

Which is a useful control. When a task list shows no middle band, that is the data saying there is nothing to build, and it is rare enough to be worth noticing when it happens.

What is missing from this chapter

There is no data here on what anybody actually does with AI. A dataset that appears to measure it was removed from this report because its meaning could not be stated. The reasoning is on the limits page.

The exposure ratings cover all 323 tasks in these 16 occupations with no gaps, and were made in 2023 against the model capabilities of that year. Read every figure as a floor.


Exposure ratings from Eloundou, Manning, Mishkin and Rock, GPTs are GPTs: Labor market impact potential of LLMs, Science 384, 1306–1308, 2024, used under the MIT licence. This page includes information from the O*NET Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA), used under the CC BY 4.0 license. O*NET® is a trademark of USDOL/ETA. People Team AI has modified all or some of this information. USDOL/ETA has not approved, endorsed, or tested these modifications. Employment figures from the US Bureau of Labor Statistics.