How can you design AI to work with procedural generation?

1 Define your goals

Before you start designing your AI, you need to have a clear idea of what you want them to do, and how they fit into your game design. What is the role of your AI agents in the game? Are they enemies, allies, or neutral? How do they interact with the player and the environment? How do they affect the gameplay and the narrative? These questions will help you define your goals and constraints for your AI design.

2 Choose your methods

Designing AI for procedural generation requires different methods and tools. Finite state machines are simple systems that define the behavior of AI agents based on a set of states and transitions. However, they can be rigid and predictable. Behavior trees are hierarchical structures that allow for more flexibility and modularity, but can be complex and difficult to debug. Utility systems are systems that define the behavior of AI agents based on a set of actions and scores, making them more adaptive and realistic. Finally, machine learning is a technique that trains AI agents to learn from data and experience, creating agents that are dynamic and intelligent, but costly and risky.

3 Test and iterate

When designing and developing an AI, testing and iterating your methods and tools is essential to ensure that it works well with procedural generation. You need to consider factors like variety, consistency, difficulty, and performance. Variety covers how diverse and interesting the environments are. Consistency looks at how coherent the environments are. Difficulty focuses on how challenging the environments are. Performance looks at how efficient and stable the environments are. To test and iterate your AI design, tools such as debuggers, profilers, and analytics can be used. Additionally, getting feedback from playtesters and users can help improve your AI design.

4 Learn from examples

Designing AI for procedural generation can be greatly improved by learning from games that have done it well. For example, Spelunky is a roguelike platformer that uses procedural generation to create random levels with traps, enemies, and secrets. The AI of the enemies is simple but effective, such as bats that fly towards the player, spiders that drop from the ceiling, and shopkeepers that chase and shoot the player if they steal or damage their goods. In Left 4 Dead, a cooperative shooter, the zombies have different types and behaviors, such as common infected that swarm the player, special infected that have unique abilities, and the witch that attacks the player if they disturb her. There is also an AI director that adjusts the spawn and intensity of the zombies based on the player's actions and the level's layout. Lastly, No Man's Sky is an exploration game that uses procedural generation to create a vast universe with planets, creatures, and resources. The creatures have varied and realistic AI that reflects their appearance, habitat, and personality. Additionally, there is an AI system that generates quests and events based on the player's location and progress.

5 Have fun

Finally, the most important tip for designing AI for procedural generation is to have fun. Procedural generation is a powerful and creative technique that can open up new possibilities and experiences for your game. Your AI design should complement and enhance your procedural generation, not limit or undermine it. Experiment, explore, and enjoy the process of creating AI that works with procedural generation.

6 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?