The concept of communicating with a computer via natural language first became popular as far back as Star Trek (1966 to 1969). In the series, we can often see Kirk, Scotty, and the gang issuing commands to the computer. Since then, many attempts have been made to build chatbots that can converse naturally with a human. During this often unsuccessful journey over the years, several linguistic methods have been developed. These methods are often grouped together and referred to as natural language processing, or NLP. Now, NLP still is the foundation for most chatbots, including the deep learning variety we will get to shortly.

We often group conversational agents by purpose or task. Currently, we categorize chatbots into two main types:

• Goal-oriented: These bots are the kind Kirk would use or the ones you likely communicate with on a daily basis, and a good example is Siri or Alexa.
• General conversationalist: These chatbots are designed to converse with people regarding a wide range of topics, and a good example would be Microsoft Tay. Unfortunately, the Tay bot was perhaps a little too impressionable and picked up bad language, much like a two-year-old does.

Gaming is certainly no stranger to chatbots, and attempts have been made to use both forms with varying success. While you may think goal-oriented bots make perfect sense, in reality the vocal/text is too slow and tedious for most repetitive gaming tasks. Even simple vocal commands (grunts or groans) are just too slow, at least currently. Therefore, we will look at the often under utilized conversational chatbots and how they can be used in gaming.

The following is a summary of the gaming tasks these bots could undertake:

• Non-player characters (NPCs): This is an obvious first choice. NPCs are often scripted and become repetitive. How about an NPC that can converse naturally about a topic, perhaps revealing information when the right combination of words or phrases are used? The possibilities are endless here, and some NLP is already used in gaming for this matter.
• Player character: How about a game where you could converse with yourself? Perhaps the character has amnesia and is trying to remember information or learn a backstory.
• Promotion/hints: Perhaps as a way to promote your game, you build a bot that can hint at how to complete some difficult tasks or just as a way to talk about your game.
• MMO virtual character: What if, while you were away from your favorite MMO game, your character stayed in the game, unable to do actions, but still able to converse as you? This is the example we will look at in this chapter, and we will get to the action part later, when we explore reinforcement learning.

There are likely dozens more uses that will evolve over time, but for now the preceding list should give you some great ideas regarding how to use chatbots in gaming. In the next section, we get into the background of what makes a conversationalist bot.