In March 2018, CaliBurger switched on Flippy, a robotic fry-station arm, with considerable fanfare. It was furloughed after one day. The robot itself worked. The kitchen around it didn't: there weren't enough people left to keep the patties prepped fast enough to match it, and the whole line's pace fell apart. Flippy came back two months later, running a reduced lunch shift, folded back into a kitchen built around people rather than in place of them.

Chefs are rated Low risk, with a horizon toward 2055, one of the most protected classifications AI Doomsday tracks. That protection isn't about chefs being harder to automate than baristas or bartenders, covered elsewhere in this series at High risk. It's that the job itself sits one level above the work robots are actually good at. Frying, chopping, and assembling are executional tasks. Deciding what goes on the menu, building a kitchen team, and being the reason a restaurant has an identity are not, and that distinction has already been tested commercially, more than once, with a clear result.

Eatsa tried to remove the human layer entirely: an automated fast-casual chain built around kiosks and self-serve cubbyholes, no visible kitchen staff, no chef persona. It shrank to two locations by 2017 and closed outright in 2019, owing back rent, before its technology was repackaged as software for other restaurants to use alongside their own kitchens. The lesson from both failures is the same: the automation works. Removing the human judgment layer around it doesn't.

Key Points

  • Chefs are rated Low risk with a horizon toward 2055, protected because the role is defined by menu design, kitchen leadership, and brand identity, not the executional cooking tasks AI and robotics increasingly handle.
  • Miso Robotics' Flippy 2 processes more than 100 fry baskets an hour, nearly double human capacity, and is piloted at chains including White Castle and Jack in the Box, while Chef Robotics' assembly systems have completed more than 70 million servings across a dozen-plus commercial kitchens.
  • Fully automated restaurant concepts have a documented failure pattern: Eatsa shrank to two locations and closed by 2019, and CaliBurger furloughed Flippy after one day in 2018 because the human kitchen around it couldn't keep pace.
  • AI menu-engineering tools like Toast IQ are now used by roughly a quarter of restaurant operators to identify underperforming dishes and optimize pricing, compressing analytical work without touching creative direction.
  • The U.S. Bureau of Labor Statistics projects 7% employment growth for chefs and head cooks through 2035, faster than average, with a median wage of $62,470, a growth profile driven by continued demand for dining out.

What a Chef Actually Does

The title covers a leadership function, not a cooking one, though it includes cooking. A chef designs the menu that defines a restaurant's identity, manages and trains a kitchen brigade, controls food cost and quality standards across every plate that leaves the kitchen, and increasingly functions as a public-facing brand, someone diners follow on social media, read about in reviews, and choose a restaurant because of. A line cook executes a recipe. A chef decides what the recipe is and why it belongs on the menu at all.

What AI Is Already Doing

The clearest automation is in high-volume, repetitive prep. Miso Robotics' Flippy 2 runs fry stations at more than 100 baskets an hour, close to double a human's typical pace, and is deployed at chains including White Castle and Jack in the Box. Chef Robotics' assembly systems, rented rather than sold, have completed more than 70 million servings across a dozen-plus commercial kitchens in the US, Canada, and Europe, with the company reporting throughput gains of two to three times and food waste reductions as high as 67%. None of these systems set a menu. They execute one a human already designed.

On the planning side, AI-driven menu engineering has become mainstream enough that roughly a quarter of restaurant operators now use tools like Toast IQ to identify underperforming dishes, restructure menu layout, and model pricing, work that used to take weeks of manual sales analysis and now runs in hours. Mezli, a fully autonomous restaurant in San Francisco built by Stanford engineers, shows what a hybrid model looks like at its most ambitious: a Michelin-trained chef designed a system capable of generating tens of thousands of Mediterranean bowl variations, executed entirely by machines, but designed by a human whose name is attached to the concept.

THE EATSA LESSON

Eatsa's technology worked as advertised: order on a kiosk, food appears in a glass cubby, no cashier, no visible kitchen. What it didn't have was a reason for anyone to come back, or a name anyone associated with the food. It shrank from a national ambition to two locations within two years and closed for good in 2019. CaliBurger's Flippy failure ran the same test from the other direction: the robot worked, but a kitchen stripped of enough human staff around it couldn't function. Both cases point at the same finding. The constraint on full automation in food service isn't whether the robot can cook. It's whether anyone still has a reason to eat there.

Where the Human Stays Irreplaceable

This is the same pattern AI Doomsday found examining art directors: as execution gets automated and cheap, the premium concentrates in curation, identity, and creative direction. Chefs have leaned into that shift publicly. Grant Achatz, chef of the three-Michelin-star restaurant Next in Chicago, disclosed using ChatGPT to help design a nine-course menu, a move that sparked real debate in the industry about where AI belongs in fine dining, but the debate itself proves the point: nobody argued the software should get the credit or set the direction. Industry commentary from chefs using generative tools for flavor pairing is consistent on this point, describing AI as useful for surfacing unexpected ingredient combinations while insisting that craftsmanship and hospitality, the actual reasons a diner chooses a restaurant, remain human work.

The Labor Market Reality

The U.S. Bureau of Labor Statistics projects 7% employment growth for chefs and head cooks through 2035, faster than the average occupation, adding roughly 14,400 positions, with a median wage of $62,470. That is a growing profession, not a contracting one, driven by continued consumer demand for eating out rather than by any automation-adjacent headwind. It stands in sharp contrast to the High- risk classifications AI Doomsday has documented for baristas and bartenders, roles in the same broad hospitality sector where the executional task is closer to the entire job rather than one layer beneath a creative one.

How to Use AI as a Chef Now

For prep and assembly: robotic systems like Flippy and Chef Robotics are worth adopting for exactly the repetitive, high-volume stations they're built for. Use the freed labor and consistency to raise the quality bar on what remains human-executed, not just to cut headcount.

For menu development: AI menu-engineering tools are a legitimate input for the analytical side, what's selling, what margin looks like, where pricing has room. Treat the output as market research, not as the menu itself.

For recipe development: generative tools are useful for surfacing ingredient pairings a human might not think to try. The judgment about whether a combination actually belongs on your menu, in your restaurant's voice, is the part that still has to be yours.

What I Think

The 2055 horizon looks right, and the Eatsa and CaliBurger stories are why. Both cases involved real, working automation technology tested against real commercial conditions, and both found the same wall: a restaurant without a human point of view attached to it doesn't give customers a reason to choose it over the one next door. That's a different and more durable kind of protection than the physical- dexterity argument that shields plumbers or electricians. It's a protection built into what a restaurant actually sells.

What I'd watch is whether that protection holds as evenly at the fast-casual and quick-service end of the industry, where brand identity is corporate rather than personal, as it does in the fine- dining examples this article leans on. A White Castle piloting Flippy doesn't have the same relationship to a named chef that a Michelin-starred kitchen does, and the executional layer is a much larger share of the total job the further you move from a tasting menu toward a drive-through window.

"The robot can fry a hundred baskets an hour without ever getting tired. It still can't decide what deserves to be on the menu, or be the reason anyone drove across town to eat it."