You got it to provide a chat response with a remotely plausible (but still somewhat unlikely, Bard probably isn’t running on MacOS servers, though training data with samples from which it might project the answer probably disproportionately is from people running code on MacOS desktops) answer to what the result of doing that would be.