The difference is that we can agree upon a set of measurements and a procedure for comparison (eg using difference of Gaussians) to determine how much of “copy” the recorded is to the original. We can repeatedly conduct this experiment (copy->measure->compare) and with high confidence we’d obtain a numerical value that can act as a “proof” it is a “copy”.

My argument is that not only that we do not know a way to conduct such category of copy->measure->compare experiments for human subjects, even with advancement in BCI, etc, it is perhaps impossible to conduct such experiment due to some nature of consciousness that we do not yet understand concerning “information”.

I used the word “mimicking” earlier as apparently when concerning “information” with humans, a (somewhat) “conscious” act has to be performed for the whole phenomenology to be interpreted as that of “copying”. We are encoding “information” in an extremely none-“traditional” way as information is studied and made sense of in computer science.

Similarly, the notion of “semantics” opens up two categorically different sets of paths for inquisition in programming language theory vs linguistics. There is something mysterious and trippy about what “meaning” and “information” really are (eg in regards to qualia).