Duration: 0:00

5 views

1

In the previous iteration, our AI-controlled enemy simply chased the player without restriction. Now, we introduce obstacles—fixed barriers that force the enemy to navigate more intelligently. This marks a crucial shift: moving from an environment with clear-cut answers to one where choices must be made within constraints.

When designing AI, a common misconception is that an unrestricted system is inherently smarter. However, intelligence is not about limitless possibility but about making optimal decisions within given limitations. This is where the No Free Lunch Theorem (NFL) comes in. The theorem states that no single strategy performs best in all possible environments. In other words, an AI optimized for one scenario will necessarily perform worse in another. This underscores the importance of defining constraints early in development.

In our case, adding obstacles forces the AI to adapt. It can no longer blindly chase the player—it must evaluate paths, avoid collisions, and, in some cases, attempt alternative routes. With this in place, our next step is to introduce a new layer of complexity: vision. Currently, the enemy has perfect knowledge of the player's location. But in reality, agents must perceive and interpret their surroundings. By limiting its field of view, we edge closer to more realistic decision-making, refining our AI step by step.