This got me thinking: given the same space, how does one create the hardest to solve maze? Obviously, you’d need to define hardest with some kind of objective metric. For AIs, this would probably be the amount of calculations performed, or time taken to solve. It would also depend on the algorithm used for solving the maze: for perfect mazes (mazes with only one solution) you could use tree search algorithms, whereas if you had loops, you would have to use graph search algorithms.
There’s some thinking about the same problem (although from a human point of view) in this Stack Overflow question: