How is the AI ?

[raindrop]

I'm really interested in this game but being a single player how is the AI ?

[LeeChard]

+1 I'm not likely to play without an effective AI

[bcgames]

List your criteria for what makes a "good AI" and I'll give you an honest answer. I wrote the scripts.

[ham17]

In a world where words like Singularity and Deep Learning are buzzing around, will we one day see a dramatic development in wargaming AI?

[bcgames]

If the Singularity is coming I wouldn't look for it here. WEGOWW2 is just a simple game engine where potential buyers might not want to make a purchase because the AI does not present them with an entertaining bang for their buck.

Bang-for-the-buck criteria--clearly stated--is still of interest when trying to answer the question--How is the AI?

[Zovs]

It seems to me that if the AI is scripted then that already give the game a leg up. I know WITE2 has both scripted and coded AI parts.

I am still on the intro scenario and learning the ropes and have not yet played against the AI properly, but seeing scripted AI is good enough for me.

[raindrop]

Does the AI
use defensive terrain
defend Victory objectives
create/escape pockets
attack with sensible odds
keep HQ's safe

or is it like the present Russian Army.

[bcgames]

Does the AI
use defensive terrain
defend Victory objectives
create/escape pockets
attack with sensible odds
keep HQ's safe

That's a good list. An autonomous AI should do all those things. I'll give you the short answer first and then the long answer once I scribe it and slap some Shakespeare on it.

The WEGO WWII "AI" is not autonomous; it does not operate on its own. So...based on the criteria listed above, here's how the game stands up for an autonomous AI requirement:

Does the AI
use defensive terrain -- No.
defend Victory objectives -- No.
create/escape pockets -- No.
attack with sensible odds -- Yes, on occasion depending on what sensible means. If you can accept 2 Para at Goose Green as sensible...then yes, on occasion.
keep HQ's safe -- Yes, on occasion but only by accident.

What we have in WEGO WWII is a semi-autonomous AI. And it does do a good job of providing the solo player a competitive game. Which brings us to the long answer...

[bcgames]

There better be more criteria than this for assessing the competence of an AI. Who is holding out? State your claim.

[benpark]

A good AI is partly on the scenario designer. I believe I saw triggers in the game editor. The use of triggers can give the impression that a group of units are escaping a pocket by placing a trigger on a map position that would make it obvious that the forces about to be pocketed would need to escape.

So far, the AI seems to do a decent enough job, so I expect the scenario design is good and tested as much as possible. I rarely play any MP games, and focus on SP. I've worked on designing things for a few games, so I know how difficult it can be to make an AI seem proficient. It's certainly worth the effort, however.

[sIg3b]

Does the AI
use defensive terrain
defend Victory objectives
create/escape pockets
attack with sensible odds
keep HQ's safe

That's a good list. An autonomous AI should do all those things. I'll give you the short answer first and then the long answer once I scribe it and slap some Shakespeare on it.

The WEGO WWII "AI" is not autonomous; it does not operate on its own. So...based on the criteria listed above, here's how the game stands up for an autonomous AI requirement:

Does the AI
use defensive terrain -- No.
defend Victory objectives -- No.
create/escape pockets -- No.
attack with sensible odds -- Yes, on occasion depending on what sensible means. If you can accept 2 Para at Goose Green as sensible...then yes, on occasion.
keep HQ's safe -- Yes, on occasion but only by accident.

What we have in WEGO WWII is a semi-autonomous AI. And it does do a good job of providing the solo player a competitive game. Which brings us to the long answer...

So where is the long answer?

[sIg3b]

There better be more criteria than this for assessing the competence of an AI. Who is holding out? State your claim.

Simple: A good AI plays to win and exploits the rules to its advantage nearly as much as a good player does (or more so, ideally).

[bcgames]

Simple: A good AI plays to win and exploits the rules to its advantage nearly as much as a good player does (or more so, ideally).

There it is.

[sIg3b]

Simple: A good AI plays to win and exploits the rules to its advantage nearly as much as a good player does (or more so, ideally).

There it is.

So we agree on the criteria for a good AI. Now the million dollar question: Does the Stalingrad AI match these criteria?

I didn´t find much in the way of reviews to answer that.

[bcgames]

It is nowhere near the ability of a good human player. It provides a good sparring partner for those learning to play the game. Humans provide the best gaming experience because they are capable of sustained strategic thought and can “feel” the skill and aggression levels of their opponents.

[sIg3b]

It is nowhere near the ability of a good human player. It provides a good sparring partner for those learning to play the game. Humans provide the best gaming experience because they are capable of sustained strategic thought and can “feel” the skill and aggression levels of their opponents.

Most importantly, an AI can´t learn from experience. That´s why I like games with many difficulty levels, so they are still replayable after the player has improved. One thing I don´t understand is this fixation on the AI not being allowed to cheat. As long as it´s not too visible and only compensates for its strategic weakness, let the AI cheat to its heart´s content!