I have many, MANY issues with LLMs, but a lot of them come down to this: I don't want my computer doing things that can't be fully tracked down in logs. LLMs and other "ai" aren't traceable, and their outputs aren't truly repeatable: their information comes from loose associative heuristics that don't follow any kind of human logic, and they are inherently prone to erroneous data. Even if there was an ai that was right 99% of the time (which is currently impossible, with all existing models being at MOST 50% accurate), I wouldn't want to use it if it couldn't show me exactly where its information came from, without errors and without hallucinations, every time. There's enough misinformation on the internet as it is. I don't want my browser, the vehicle I use to navigate information, adding more. In addition, I don't really want my bank account information stolen when I read a Reddit post, thanks.