When you fill out the "Sample Utterances" section of your interaction model, associating sample phrases with specific intents, what you're doing is giving Alexa "hints" and increasing the chance that Alexa will accurately recognize the phrases listed, and choose the correct intent.
However, it should be understood that Alexa's recognition is not limited to those sample utterances provided. A user can input an undefined nonsense utterance well outside the scope of those samples provided and Alexa will still attempt to choose an intent. This behavior is by design.
