I use it mostly to explore the information space of architectural problems, but the constant "positive engagement feedback" (first line of each generation "brilliant insight") start being deeply insincere and also false by regularly claiming "this is the mathematically best solution - ready to implement ?" only that it isn't when considering it truly.

I have moved away from using an LLM now before having figured out the specifications, otherwise it's very very risky to go down a wrong rabbit hole the LLM seduced you into via its "user engagement" training.