zlacker

Exactly. I've worked in manufacturing automation long enough to know that the equipment needed to generate the huge variety of food even a mediocre cook can prepare would fill multiple kitchens. Automation equipment that's safe and reliable tends to be very big in relation to the items it operates on.

>>zwieba+(OP)
One could imagine something like an industrial robot arm that can swap implements being able to manage a relatively large variety of recipes given some staple ingredients in a limited amount of space, but it would be tremendously complex to build and cost more than an industrial robot arm. Keeping it clean and well maintained would be a nontrivial effort, even if the arm includes programming to clean up after itself.

Probably like a billion dollars to develop the first prototype, and each copy would be in the hundreds of thousands of dollars range, maybe eeking down into the tens of thousands of dollars range if they work really well and become inexplicably popular. Even then it would be up to the customer to keep it stocked with ingredients in a specially designed containers in the included pantry and refrigerator.

It's really the same reason McDonalds never really went through with that fully automated restaurant threat. A person can do the same job for minimum wage, so the robot will never be cost efficient unless someone else does all of the R&D for you, and even then it's highly dependent on being low maintenance.

>>jandre+x5
Articulated robot arm sounds like the most flexible solution. Trickiest part here would be safety - any robot with the power to chop, grind, cut would also have to have a good safety system or be enclosed.

Right now I'm thinking about tasks I don't enjoy: chopping onions, peeling carrots or potatoes, anything where I have to touch meat. All those would require very advanced sensing. Come to think it, that last one brings up the important topic of food safety and sanitation, the whole thing would have to be able to withstand a washdown.

>>zwieba+na
Not to mention the risk of fire if something goes wrong in the cooking process. (Or, less directly, the risk of getting sick from improperly-cooked food.)

>>jandre+x5
I'm thinking here of ill-fated robot kitchen startup Zume's pizza robots. You can probably still find videos of them in operation on YouTube. The things that most struck me when I watched one were how slow they are at what they do, how much space they took up, and how much human intervention they required to operate.

>>drewze+eq
If I wasn’t so tired and literally dropping off to sleep right now, I could make a decent argument that cooking requires AGI. Driving in our complex urban environments too for that matter.

>>zwieba+(OP)
Nothing like an experienced engineer to slow down progress. Give me a hungry engineer right out of school and we will get this thing done. Sure we may injure thousands of people in the process but we can deal with that after the product is on the market; oh yeah and I will blame the engineers for it too.

>>zwieba+(OP)
The thermomix seems the closest thing I can think of; automatic stir-while-heating covers a fair bit.

>>jandre+x5
The robot would need to operate on ingredients that are within a narrow band of parameters. A cook will adjust the receipe if for example some vegetables or fruits are particularly ripe or tough or whatever, while the machine would have an incredibly hard time doing that. Realistically, the ingredients would need to be standardized (within a narrow band), which would increase the cost of final meal. Of course, fast food is practically doing that already, but in result they can only do simple meals with mediocre results.

>>burnto+uv1
I would guess that this is why cheap packaged foods tend to be ultraprocessed. The machines need consistency to operate reliably, and the easiest way to make food consistent is to turn it into a liquid, powder, or mush.

>>jamiek+Nm1
Adjusted Gross Income? (Not meaning to be obtuse, but I'm not familiar with the acronym in this context.)

>>drewze+KP2
Artificial General Intelligence, or "strong" AI. A lot of the things we consider AI today is often called "weak" or "narrow" AI, in that its scoped to very specific tasks. Hotdog/Not Hotdog kind of things. General intelligence is more like our kind of intelligence, where it is adaptable to a wide variety of tasks without necessarily needing an entirely different model or structure.

https://en.wikipedia.org/wiki/Artificial_general_intelligenc...