The AI is actually very primitive for navigating mazes. It is the same AI for every character and does not vary with ride tastes. All the AI does is prefer walk

The AI is actually very primitive for navigating mazes. It is the same AI for every character and does not vary with ride tastes. All the AI does is prefer walking in straight lines and when it reaches a forced decision it takes either direction with equal rating. It does not have any idea where the exit is or do anything fancy. It can easily end up going round the same area on repeat. Source I read the x86 assembly.