Explain Invalid or "don't care" inputs in Boolean Algebra

As far as practical usage in truth tables/kmaps, I understand perfectly that they can be either 0 or 1 at the engineer’s whim in order to create the most simplified expression. What I am confused about is when or what exact set of criteria would cause an input to be considered an invalid input.

in the lesson, it says the following

“There are certain conditions where a function may not be completely specified, meaning there may be some inputs that are undefined for the function”

what exactly are these certain conditions? they seem to vary depending on the function. Is this something we will cover later on?

These would be inputs which, for the purposes of your function, would never occur; because the point of the kmap is to build out the simplest function (for the sake of the simplest circuitry), eliminating impossible inputs simplifies your problem.

Imagine you are building a circuit whose output is to turn an air conditioning on and off in your home. You want to conserve power, so you want the A/C to only turn on only if the temperature is above a certain threshold, there is at least one person present in your house, and all the doors are closed (that way the cold air doesn’t escape outside). Now, let’s imagine that you are a person who is rather protective of your home, so you are never going to leave your door open if there isn’t somebody inside. Thus, for your use case, all inputs in which “someone is home” is false and “a door is open” is true are irrelevant for your circuit as they will never occur. Therefore, you should flip those impossible outputs to true or false depending on what makes the simplest circuit.

From my understanding, in all simple boolean expressions (which is what we are working with here), inputs will only be invalid based on the context in which you are using the circuit: what is your use case and, in that use case, what things can we rule out?