Calling AI things like “smart” or saying it “knows” something might sound harmless, but it can quietly mislead people about what AI actually does. A new study shows that news writers are more careful than expected, rarely using strongly human-like language. When they do, it often falls on a spectrum