The Course Work and Utility

• So how might you use this stuff for the course work.
• You can't use minimax for it as it's not adversarial.
• You could however use a game tree with the probability of the cards as likelihoods you would move to a particular state.
• This is not unlike a bayesian tree.
• You could also use goal trees (and/or trees) to see how goals could be resolved.
• This is just the underlying agent. In the simplest case, the agent could just keep track of the likelihood of particular cards (from both the player and infection deck) are going to come in this turn or the next.
• The agent can interact with the conversation in an interesting way.
• As the agent becomes more sophisticated, it can have a more sophisticated conversation.