The article captures what is evident for many language model users. But it's good that it's captured. I have two more thoughts. One is that many people have pl

The article captures what is evident for many language model users. But it's good that it's captured. I have two more thoughts. One is that many people have plausible but not true knowledge of things. They often share it online. As I have become more and more specialized in my tech niche, I've started noticing that online advice and discussions of that niche involve only cocktail party-level knowledge, which is wrong in many cases. The other thought is that our memories also work in an approximate way. So "plausible, but not necessarily true" aligns with our thinking. Flooding the internet with convincing but not wrong "facts" could be very toxic to our minds, given how comfortable they are with that type of information.