My guess is that if you put a heater into a volume of water, it'll work flat out to heat that water until it reaches temperature where it'll settle down into some kind of equilibrium. with energy in from the circ, energy out via radiation, conduction, convection, that kind of stuff.
If there is a lump of frozen food in there then the power consumption profile will be different. It'll look like it's heating a much bigger volume of water or in a much colder room.
It'd be fun to get access to the machine state to do some experiments with what you can tell from that data.
Maybe you could make a predictive model based on how much the power consumption changes when you dump something into the water. It might be able to give approx cooking times?
Anyway, fun thought experiment.
