The Zebra Puzzle is a well-known constraint satisfaction problem that ChatGPT knows about. But can ChatGPT solve the puzzle or problems like it step by step in accordance with the rules of logic? The aim of this project is to investigate how far ChatGPT is able to reason about such problems, and how ideas such as Chain of Thought can be used to help it out. Data structures related to finite-state methods will be used to record the reasoning about such puzzles.