AI in wargaming

[Asberdies]

i guess most poeple have heard bout generative ai like Chapgpt or other ai engine by now. I keep hearing about the revolution generative ai is bringing to us and i am wondering if poeple bagan to think about using those ai to A Last get a real ai in our wargame. That would be a revolution!
What do poeple and mostly game designer think about this?

[mainsworthy]

RPG coversations would be good, I think wargaming can be like chess so we already use A.I. but i know your saying you want the new shiny ai to fit

[RangerJoe]

Those AIs are often wrong, especially in the legal world.

[Zeckke]

Fear all AIs; (will substitute de human when incorporated where a few people Know) for example i had fear sometimes playing the AI of Strategic comand also on uncomon Valour (massive agressive bombardements) and WITP is the correct AI

is just another step for robots to oput into his head of you want to debate.

i just saw for asecond HERE, where..impressive,

[gamer78]

For narrative they use AI in space strategy games for races background. And sometimes for graphics. Wargaming is not -should not be-really competitive gaming among players. It is more about historical interest about different conflicts. But better AI needed for these complex systems for more popularity. EX: SGS Korean War and Taipings, pitty that those should have better AI.

Edit: Traditional wargaming does suffer from AI but more about same conflicts over and over again, for their market they produce games. New Strategic Command beta is no surprise with Pacific War, as the engine is best for WW1. That's why SGS games as example -as innovative and brave- they have different conflicts such as AGEOD did in the past but unfortunately low profile AI. That is pity for expectations.

[pzgndr]

I am amused. Like, sure, I can just plug in ChatGPT to generate a challenging computer opponent for Empires in Arms. I've been working on the AI for years to help make it more challenging. Lately, I've been working on grand strategy options for the AI. Absolutely none of this stuff is trivial. If it was easy, anybody could do it. I literally LOL whenever I read about AI hype in the news. When I stop laughing, I get back to code development...

[OldSarge]

How about a nice game of chess?

wargames.jpg (14.41 KiB) Viewed 737 times

[gamer78]

Napoleon whole campaign is not good example for AI. I didn't see any game apart from -boardgame people with house rules- has any success with it's design with full function AI or played commonly in pbem or multiplayer excluding Total War in PC . If that was common I would push Celali rebellions during 16Th 17 century next title. It has also importance.

[sPzAbt653]

I would not like to play my favorite game against Hal, or Walter/David or Ava. Therefore I don't think AI is good for gaming. Working on Programmed Opponents seems to be a better pursuit.

[RFalvo69]

Recently something happened in the world of AI and gaming that it is not discussed enough.

Rio Grande Games put out a digital version of their card game game "Dominion" - a medieval themed game which only needs the base box to be played but that was so succesful to generate 15 expansions. The AI is based on an advanced neural network, and on "Hard" level it just obliterates any human player.

"Dominion" is a game I know quite well. In past times I played the tabletop version a lot with my daughters, friends and relatives because it is easy to learn, hard to master and full of flavour. It is also very fast, with a game lasting, usuially, 15-20 mins. For years was our family game for Christmas gaterings and such (along with Dixit).

All the above just to say that I thought how to play "Dominion". But if I look at my stats on the App I recorded 90 games with 4 wins.

The story of how this AI was developed is really interesting and can be found here. It is a fascinating read, especially the part about how they wanted to create a AI capable to learn the game rules and apply them - a must if you want to ship the game with 15 expansions who add different rules to the base game without rewriting the code everytime an expansion is added.

As I said, I don't think that the implications of this breakthrough are fully realised. One could point out how this AI was built around a specific genre: the competitive card game. The programmers mention how there was already an earlier strong AI capable of playing Starcraft, but only "on certain maps with specific races". I now wonder if the "lessons learned" here can be applied to strategy games - and, by implicatin, to new AIs for wargames.

[PoorOldSpike]

The AI in 'The Troop' is one of the toughest i've ever gone up against in my 41 years of computer wargaming, it's so bloody good that some players swear it's cheating, but the devs assure us the AI plays to exactly the same rules as the human player..

[gamer78]

I would not like to play my favorite game against Hal, or Walter/David or Ava. Therefore I don't think AI is good for gaming. Working on Programmed Opponents seems to be a better pursuit.

From my readings and understanding about gaming industry: year 2024 will be worse than 2023 for profits. Programmed Opponents pursuit already started as early with desktop Pc gaming with original designers and modders. Perhaps many games PO improved a lot with volunteer modders not looking for finance. Whole Napoleonic period is not good example I mentioned before was about ,it is not decided yet what the whole campaign should be among pbem players at least in PC gaming.