[As an aside, is the GPT feedback functionality correct? For example, if the secret word is "peals" and the user's guess is "apple" the feedback is going to be "a?p?p?l+e?". However, I would expect the feedback to be "a?p?_l+e?". Because the first 'p' exists in the secret but is in the wrong place and the second 'p' does not exist in the secret word.]