I don’t think that’s a good way to measure whether AI thinks or not. Many humans might make that mistake but you wouldn’t then come to the conclusion that that person can’t think