We know how ML models work: A machine-learning model finds patterns in the input data, looks at its vast statistical knowledge (lossy-encoded as multi-dimension

We know how ML models work: A machine-learning model finds patterns in the input data, looks at its vast statistical knowledge (lossy-encoded as multi-dimensional latent space), and then generates patterns which are statistically close to the input data. If this instance of 'statistically close' corresponds very closely to what humans perceive as 'semantically close', we perceive this as parroting. If this instance is somewhat distant but still connected to what humans perceive as 'semantically close', we perceive this as creativity. If this instance of 'statistically close' is very different from what humans perceive as 'semantically close', we perceive this as an error, a glitch, a fluke, or the model not working well. This is complicated by two things: (a) we have a very limited understanding of what 'semantically close' means. (b) we have a very limited understanding of how human brains work. Edit: formatting