TheRavenHeart wrote:
Speaking as a software engineer, if Playdom could cost effectively improve the AI massively beyond it's current capabilities, they most certainly would, as it would please a large chunk of their audience.
However, the sheer number of variables involved in factoring in EVERY possible scenario would be difficult at best if there were only 6 heroes available to use for PvP with no alternate costumes and no ability to equip ISOs due to the fact that you would still have 15 possible combinations of hereos and 4 moves for each hero. This isn't even factoring in the fact that you would still want an AI that would respond differently depending on the different classes of heroes facing each other and various HP and SP levels for each character. This is if there were only 6 heroes in the game, no ISO socketing, and without factoring in the agent.
With all of the various equipment you can load on the agent, you aren't looking at an infinite number of possibilities, but you are looking at an arbitrarily large number of permutations that to the human mind, may as well be without an upper bound. Though not as large, now thanks to Reactive ISOs once you get over level 125, you have a large number of permutations, that while not as high as the number of permutations possible for the agent loadout, still gets multiplied by three since both heroes and the agent have them.
People seem to think that building an AI that could handle this is easy, and that any AI should be able to handle the complex process that the human brain can deal with. If building the perfect AI was so easy, the machines would have already replaced us.
Getting a little nerdy -and since I'm doing my masters on artificial intelligence, I can get pretty nerdy on this- it wouldn't be that hard to build such AI, the problem is that they're probably using scripting. The answer is in decision trees, which I can clearly tell that they are not using...