Jump to content

AI Improvements from Demo


Recommended Posts

Can any of the beta testers out there tell me if they have noticed a significant improvement in the AI from the beta demo. Not to brag, but I have been able to win against the computer rather easily. I've now played each scenario from both sides and haven't really had a close match yet. Maybe I've just been lucky.

I'm assuming that the overt bugs have been fixed (such as platoon leaders making solo charges into heavily fortified positions). But I have the impression that the AI in the demo doesn't make very intelligent tactical or strategic decisions. It tends to keep moving units into areas where I have just decimiated other units. In short, it doesn't seem to adapt very well to prior events which would definitely affect a human player's decisions.

By way of example, I'm currently playing CE as the US. I have three shermans on the wooded slope which is near the vertex of the triangle formed by the wheatfield and the church. They have a very narrow corridor with a clear line of sight to the German edge of the map. The AI moved one Stug into that corridor for three consecutive turns. Each turn the offending stug was promptly dipatched. By the end of Turn 6, I had destroyed all three Stugs, which fired a combined total of 2 rounds (neither of which hit). I get the impression that if the AI had 10 Stugs, it would keep moving them though this area one at a time and I would keep blowing them to pieces. Shouldn't the AI have figured out that maybe it wasn't such a good idea to keep moving its Stugs into this particular spot, especially after the first two exploded? The AI clearly knew that my Sherman's were there and that they had a clear line of sight since there was german infanty in the same area as the destroyed tanks. As you can imagine, the rest of the scenario hasn't been much of a challenge, with 5 shermans against the unsupported German infantry.

I realize that there are limits to what one can expect from an AI opponent. Although the AI in the beta demo is better than the AI in most commerical products, I am still hoping that the final product makes significant improvements. I'm curious to know what the beta tester's experience with the AI has been in more recent builds.

Link to comment
Share on other sites

Guest Big Time Software

Hehe... covered several times in depth. Just keep in mind that when you replay a scenario the AI plays it for the first time since it has no memory. So you *should* win just about each time you play after the first. Also, there is a bug in the demo concerning the victory ratings, which are FAR too generous. So don't use the rating as a guide for how well you did.

Do a Search on this BBS for "StratAI" or something like that. There were several long, detailed threads dedicated to this topic within a few weeks of the demo's release.

There have been significant improvements for sure, but most of what you mentioned you are looking for in terms of "intelligence" is beyond the capacity of any game for many years to come. This is covered very well in previous posts too.

Steve

Link to comment
Share on other sites

Although the AI does some silly things that most human opponents wouldn't, Steve makes a good point when he says the main advantage a human player has is knowledge of the scenario.

My first time playing Last Defense as Germans, the arrival of the M18s took me completely off guard, leading to burning tanks and halftracks everywhere. The second time the Tiger was positioned to take on the M18s as they arrived, and the halftracks were all hiding behind woods/buildings by turn 10. If you really want to cheat you can target arty on the position to come down just as the M18s arrive. What AI can match that?

Link to comment
Share on other sites

The scenario designer has some effect on the replayability factor, wherein the AI is not too badly at a disadvantage. Careful map design, randomizing the appearance of reinforcements and original placements as to time and place have a great deal to do with keeping the human off balance in replay. Also the forces available to both sides offer some opportunity to balance the difference.

However, with too much difference here, a first play win can be made virtually impossible. Such a thing may satisfy those dedicated to replay after replay working out the fine details as in a puzzle, but something that sits sourly on the stomach of those warriors, who would rather play in a battlefield rather in an arena against the mentality of the designer. The force of the designer's personality in this case can be felt in the scenario as a stronger element than than the scenario itself. Let the gods of war act so as not to be intruders, but so as to enable the illusion that the players are independent and agents of free will only subject to the fates of the moment.

Link to comment
Share on other sites

×
×
  • Create New...