It's always been in fairly low stakes corporate settings so not a major worry but I do think people don't recognise how error-prone it is

I'm alarmed by how many people in professional settings take it as gospel "well chat gpt says X" used to close down discussions etc.