Artificial Intelligence(AI) is a term banded around most game systems and is a general bucket for several techniques for machine-based learning systems. Its sole aim is to fool the user/player into believing that the system is behaving like any living being, mainly to challenge the player in head-to-head battles or helpful supporting characters.
Some systems used to achieve this are as follows:
- Path-finding: This helps AI-controlled entities navigate through levels to a specific destination
- Flocking: This orders how multiple AI entities will relate to each other within a given area
- State machines: These are fixed and basic sensor-driven intelligence to drive AI actions.
- Rule-based expert systems: These are the defined logic systems for an AI entity to derive action from and aid decision making
- Neural networks: These are advanced learning networks for AI entities, typically used to predict the performance of the AI and also understand the predictable behavior of opponents
- AI algorithms (reinforced learning / simulated annealing / genetic calculations): These are many different ways to reinforce neural networks and decision engines for better predictable behavior
The area of AI can be a very complicated minefield. It is often seen by some as a "nice thing to have"; however, getting it right is a very long and drawn out task no matter the size of the project.
My advice, especially if you are just starting out, is to lean on existing implementations either through the asset store or the Unity wiki to begin with and learn from there. The whole subject of AI has spawned numerous books and entire sites such as http://aigamedev.com/ (a fantastic general resource).
Start simple and move from there. For the purposes of this book and the RPG game, we will focus on a simple state machine implementation using basic sensors to help drive the AI.