Results 1 to 30 of 40

Thread: Future AI

Threaded View

Previous Post Previous Post   Next Post Next Post
  1. #11
    Member Member Didz's Avatar
    Join Date
    Oct 2000
    Location
    Bedfordshire UK
    Posts
    2,368

    Default Re: Future AI

    The key pre-requisites for an effective AI are 'Situational Awareness' and 'Behaviour Modification'. Chess programs do quite well only because the improvements in processor technology now allow them the assess the possible consequences of every potential move they can make virtually to the end of the game and then choose the option that gives them the best chance of success. This gives chess programs a massive advantage in situational awareness that most human players could never match. However, in most other games the AI's level of situational awareness is extremely limited and most AI's are restricted to very simple reaction triggers.

    In my current MTW2 Turkish campaign for example the Hungarians were beseiging the Turkish held fortresses of Bran and Sophia, and whilst their two main armies were busy I slipped another army between them to assault the lightly held Budapest. This army was fully visible to the Hungarian AI, and even drove off a few small Hungarian armies during its advance, but it was not until it was within one turn of Budapest that the Hungarian AI's situational awareness noted this army as a threat and it immediately lifted the seiges of both Bran and Sophia to try and save its capital and faction leader. Much, much too late, especially as my army had trebuchets, however, the AI simpy failed to register the threat at all until it was too late suggesting that there is no situational awareness routine that thinks ahead of the immediate snapshot taken during the current turn.

    Likewise, very few AI's currently have the ability to modify their behaviour over time. To be really challenging an AI needs to constantly monitor its own performance and that of its opponents and test alternative strategies to improve its game. At present we see the AI in MTW2 constantly repeating the same losing strategies that it tried last time and repeatedly putting together army compositions that are pre-destined to fail against our own. A human opponent would learn from its mistakes and at least try something different so AI routines really need to have some system for monitoring their own performance over time and seeking to improve.
    Last edited by Didz; 07-12-2007 at 17:00.
    Didz
    Fortis balore et armis

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
Single Sign On provided by vBSSO