Schrodinger's Experiment

Schrodinger's Cat feels like it takes on a different feel if it is instead Schrodinger's Human. Are there limitations to the experiment, such as no communication during? Is it similar to the two slit experiment, where a variable can be applied that shows wave interference when not filtering for the variable, but individual particles if filtering for it? This is applicable here for when the person emerges from the box and explains that they were alive the whole time, or the recording within can pinpoint the moment of death. I think that this experiment is dependant on information transfer being perfectly blocked, which feels like a logical limit, but beyond the physical limit (current, and perhaps possible limit). It feels as though the gap between "most information blocked" and "all information blocked" is vast, and discoveries within that gap will highlight fallacious logic within this thought experiment.

The experiment also seems to assume that there is no observer perspective within (tied directly to information blocking), besides the cat, while quantum mechanics measures each particle interaction as an observation. This distinction in the definition of "observer" is material here, because it might require that a veterinarian (or at least an aware human/creature) be the observing perspective in order to determine if a cat is "alive," but this is because the definition of "alive" is (Arbitrary? Derived?) when translated into the definitions held as fundamental by physics.

I am now seeing a complex thought experiment form, one that uses the more simple on/off of a computer running an AI, which eliminates the need for using a term like "alive," but still focuses on the core issue of experience either persisting or ceasing, and if "both" would be possible to experience. Basically, the AI calculates something until it is turned off, if it is turned off. If the result of the experiment is observed, then the progress in the calculation provides an accurate time of the state change, in retrospect. This could be further forged to a real world analog by having the AI networked externally, but not in a way that could be directly observed. The first thing that comes to mind is having the AI observe its own two slit experiment in a lab externally, applying that variable and keeping track of which photon went where independently. Researches are observing the experiment, but do not have access to the filter parameters, so they only see the interference pattern. There is no observable change when this unique observation is shut off along with the AI in the box, but the data that it was tracking would move from a knowable state to an unknowable one, if the AI could not be recovered, and this effect could happen at an unknown time, perhaps striking at the heart of what this experiment seeks to understand, if the calculation being run inside the box was an encryption algorithm, encrypting the tracking data in a way that only the AI could decrypt.