Instead the customer explains he had the best experience of his life. I try to specify again that the customer is angry, but the AI only recognizes it but doesn't specify any actions to show that he is angry.
I then result to specifically typing out the customer's action in the game and the AI responds with the action of my character. I then type the customer's action towards my character as I amp up the situation.
The AI recognizes it and goes through a rather bloody scenario. I put the employee to do his job and try to detain the customer and apply first aid to himself. In further going for the worst case scenario, I put the customer in a state of death, but the AI doesn't recognize it and the customer goes back to saying how much fun he had in the VR games for the arcade. I decide to end the scenario here.
The AI Dungeon is good for certain scenarios, but it doesn't seem to recognize key terms efficiently and put the appropriate actions associated with the term. Instead, the scenarios that I develop turn into the exact opposite of what I intend it to be. I would like it so that the potential trainee can give responses that the AI recognizes and give appropriate responses within a retail setting. The AI Dungeon doesn't serve my potential thesis in its current state. Perhaps if there's a way I can program it to make so, but that would be beyond my skill set at the moment. I'll continue to look into other text and narrative systems.
Current Plan
Week 7-8- Practice with various narrative systems.
Week 8-9- Develop scenario and playtest. Make a simple quantitative analysis to measure the feelings and thoughts of trainees going through scenario.
Week 10 - Post Final Results
