Draw a circle on your computer. Did you do it? Are you sure?? Zoom in 9n the circle. Still look round? Ok...

AI and wfca are sometimes used in tandem to build realistic conversations. For example, a trained Chat AI can produce a reasonable statement like "The sky is blue. People have blue eyes. Therefore people's eyes are the sky." It's a pretty lackluster line of logic a trained AI proposes but a series of probabilistic adjacency rules such as just one that says "Human anatomy and weather are mutually exclusive" or something to that effect can dramatically clean up the output given by the AI. If you know anything about wfca, this is exactly how the algorithm works where a set of adjacency rules control the final output based on what already exists. Hopefully that clears it up a bit.

TLDR; AI is often very far from perfect. Algorithms like wfca can help get it quite a bit closer.

WFCA is at the core of alot of ML approaches to synthesize effective NLP in chat AI.

Look up the "Wave Function Collapse Algorithm". We've been using AI (more accurately machine learning) for a long time to quickly build highly complex game worlds.