For my final I decided to combine most of my past homework in one. Because I realized that on their own, they are not that special & meaningful, they are not subverting the current relationship that exist between Voice Assistants & people. But by combining them I get a chance to show a different more balance relationship.
The user has to do something for the voice assistant first - spend time on the food puzzle. Only then they can access the voice assistant. In return, voice assistant can give a user a snack telling, a prediction of what the user gonna eat today. Hence, the user doesn't have to decide anymore and as a result regret that they ate because the responsibility for that is on the voice assistant now. The prediction is given based on how good/bad the food was for that user that day (sentiment analysis), and there are around 8 different predictions.
Puzzle and prediction options
Then the Snack Teller tells the user a joke even if they don't want it. So the user has to listen as a sort of 'thank you' for the prediction. After the joke, the Snack Teller feels tired and ends the conversation with the user which reinforces the ides of the voice assistant being its own master.
Video of interaction with the Snack Teller
I still have problems with training it because no matter how many phrase I added for food evaluation, sometimes it still goes to fallback. I also need to change the phrasing a bit, as Nicole suggested, so the relationship between the user & voice assistant would be more clear.