I think there's a big gulf between people who use e.g. coding agents (where this would be actively detracting + annoying) and people whose exposure to LLMs is mostly google search AI overview (where the output *is* ~equivalent to the average of what people are saying)
