Chatbots, or conversational agents, are an exploding trend in AI and are seen as the next human interface with the computer. From Siri, Alexa, and Google Home, there has been an explosion of commercial growth in this area, and you most likely already have interfaced with a computer in this manner. Therefore, it only seems natural that we cover how to build conversational agents for games. For our purposes, however, we are going to look at the class of bots called neural conversational agents. Their name follows from the fact that they are developed with neural networks. Now, chatbots don't have to just chat; we will also look at other ways conversational bots can be used in gaming.

In this chapter, we learn how to build neural conversational agents and how to apply these techniques to games. The following is a summary of the main topics we will cover:

• Neural conversational agents
• Sequence-to-sequence learning
• DeepPavlov
• Building the bot server
• Running the bot in Unity
• Exercises

We will now start building more practical real-world working examples of the projects. While not all of your training is complete, it is time we started to build pieces you can use. This means we will begin working with Unity in this chapter and things may start to get complicated quickly. Just remember to take your time and, if you need to, go over the material a few times. Again, the exercises at the end of the chapter are an excellent resource for additional learning.

In the next section, we explore the basics of neural conversational agents.