I was playing Chinese against Delhi yesterday after the patch, and I found out that when its elephants are attacking my buildings, the attack speed is much faster than 1 attack per 5 seconds and my buildings are absolutely melting. The behavior of the elephants also look like alternating between attacked and moving to the side for a split second.
I found it amazing that the AI seems to really learn from the way the human players plays (if the hardest AI is actually ML-trained) or they learn this trick when training with each other. I don’t think pre-programmed AIs would display this kind of behavior.
1 Like
Haha. Maybe the devs are controlling these elephants 
Its possibly fluke where the AI is constantly changing its mind about moving the elephants or attacking the building so its unintentionally doing the cancel exploit.
Or its learning and we should all be afraid!
1 Like
I’ve seen similar behavior from computer. Seems it can’t make up its mind sometimes and looks like it’s using cancel exploit, though, usually not in its favor.
I hope it is learning because as an RTS veteran, despite 0 AoE experience, I must say the hardest AI is a joke. Played with 2 friends who had the same experience versus 5 hardest AI and that was actually incredibly easy. The main issue? Rolling siege over the huge map to wonders because all AIs roughly build wonders at the same time. It was more a rat race than anything.
My guess is that hardest AI roughly equals 1.000 elo, tops?
From my experience, it’s more like 700 ELO in this game, if you send 1 military unit to their town at the start. Maybe 1000 ELO if you left them build up a lot.
They really need to improve the AIs resolve to not give up on attacking. It will so often have enough units to do big damage, but then simply turns its army around. Very dumb AI.
I’ve had the AI cut the woodline by a wall to get their army through so it could definitely be learning.
1 Like