I believe that this is simply because of the way we train ML, with labelled data. It is quite conceivable that we could get an ML model to recognise cats just by some form of multidimensional clustering of training data.
This would also impact clustering.
That said, I think even for humans there's a similar issue: we spent millennia clustering things into groups and labelling those groups, which is why the Catholic church had rules about no meat on Good Friday but fish was fine and beavers counted as fish (and there is now a podcast titled around the idea there is no such thing as a fish*). For cats, I don't see it myself but the fossa is described as "cat-like".
* https://en.wikipedia.org/wiki/No_Such_Thing_as_a_Fish#Title
Of course what a cat _is_ to me is not what a cat _is_ to you, because we necessarily have different memories of interactions with cat-like beings. If you show some babies a cat for the first time, they'll necessarily see it from different viewing perspectives. Even if you put VR glasses on them and show the exact same video, they'll have different contexts: "I first saw a cat when I was sitting next to my friend", "I first saw a cat when I was thinking of ice cream", etc.
But they all saw the same cat, they'll see many other cats, who are all similar. So everyone will understand that "things like these are cats", but everyone will have their own understanding of a "cat" because their memory is different.
Heidegger best revealed to me the limitations of supposedly "objective" thinking.
Heraclitus: "No man steps in the same river twice"