Fair points, even though I am not sure how a “control” is supposed to look like in this context. Got something better on the impact of LLM usage on the users brain, especially in regards to learning?