I tend to think this way as well, though I find team stochastic parrot is shrinking by the day. It is very difficult to explain it to people who don’t have at least some grasp of what’s going on inside an LLM, and how they’re trained. We have no psychological “immune system” for something that mirrors our ability to deliver an apparently well-reasoned response to a question in our own language. All our instincts tell us to assign it sentience and treat it as an independent actor with it’s own internal motivations and personality. Even I catch myself doubting that a pile of matrix multiplication isn’t “alive” in some way after using it for a while.