Page 1 of 1
“AI” cheating?
Posted: Tue Mar 18, 2025 4:41 pm
by ringerthrawn
So I still play AI games to pass the time between PBEM turns. I'm trying the revamped scenarios in 28b, Scenario 1 in particular, and I'm in Oct 42. So I moved over to the J side because I am curious as to what the "AI" (game engine, whatever) is doing. Just taking a peek behind the curtain. Tokyo has 14 MILLION in supplies. Fuel, supplies, oil, resources are all WAY out of parameters at many of bases. Is there something in the coding that gives the COMPUTER a boost when playing a real person?
Re: “AI” cheating?
Posted: Tue Mar 18, 2025 4:44 pm
by dr.hal
yes it does...you will also find that the AI will move ships and other things around.....all in the name of making the game more playable (?)!!!!
Re: “AI” cheating?
Posted: Tue Mar 18, 2025 4:47 pm
by ringerthrawn
Pretty sure I’ve caught the AI teleporting ships. I never grabbed a screen shot, so I don’t have proof.
Re: “AI” cheating?
Posted: Tue Mar 18, 2025 4:52 pm
by dr.hal
I've never seen it myself as I don't play against the AI but read of it over the years on this forum...
Re: “AI” cheating?
Posted: Tue Mar 18, 2025 4:58 pm
by RangerJoe
Yes, it teleports ships and other things as needed.
Re: “AI” cheating?
Posted: Tue Mar 18, 2025 11:10 pm
by Bo Rearguard
The Japanese AI also leads with its glass chin a lot, so that kind of makes up for the "cheats".
Re: “AI” cheating?
Posted: Wed Mar 19, 2025 1:04 am
by BBfanboy
Just for a parameter on AI teleportation - the units being teleported must be in a friendly base. Ships not in a port for example cannot teleport. I assume the same is true for aircraft and troops.
Re: “AI” cheating?
Posted: Wed Mar 19, 2025 1:15 am
by ringerthrawn
Bo Rearguard wrote: Tue Mar 18, 2025 11:10 pm
The Japanese AI also leads with its glass chin a lot, so that kind of makes up for the "cheats".
So true. There are times you are just like, “really? Ok, I’ll kill those ships too.”
Re: “AI” cheating?
Posted: Wed Mar 19, 2025 1:15 am
by ringerthrawn
BBfanboy wrote: Wed Mar 19, 2025 1:04 am
Just for a parameter on AI teleportation - the units being teleported must be in a friendly base. Ships not in a port for example cannot teleport. I assume the same is true for aircraft and troops.
Now that’s good to know.
Re: “AI” cheating?
Posted: Wed Mar 19, 2025 12:02 pm
by btd64
Bo Rearguard wrote: Tue Mar 18, 2025 11:10 pm
The Japanese AI also leads with its glass chin a lot, so that kind of makes up for the "cheats".
True....GP
Re: “AI” cheating?
Posted: Wed Mar 19, 2025 1:09 pm
by RangerJoe
What is really bad as the Allied player is when you surprise an IJN 2 carrier task force with four 14 inch hits from the POW on a 'kaku class carrier during the first combat round. The bad part was that the carrier did NOT sink!
Re: “AI” cheating?
Posted: Wed Mar 19, 2025 1:17 pm
by ringerthrawn
RangerJoe wrote: Wed Mar 19, 2025 1:09 pm
What is really bad as the Allied player is when you surprise an IJN 2 carrier task force with four 14 inch hits from the POW on a 'kaku class carrier during the first combat round. The bad part was that the carrier did
NOT sink!
Not frustrating at all.
Re: “AI” cheating?
Posted: Sun Mar 30, 2025 5:24 pm
by PaxMondo
So, what PBEM players call "cheating", AI players call "AI".
I play the AI.
The AI does not cheat. Period. Everything it does is for a reason, and that reason is to allow it to emulate a human player as best it can using the technology that it has.
Moving Ships. The AI cannot plan. So when a script runs and it needs ships, it collects them and uses them. Where the ship was is referred as "false Intelligence" or "maskirova" by AI players.
Supply to islands. The AI cannot create TF's for supply. Thus, it uses Hobie Cats to move supply to islands. These are completely undetectable by the opposing forces. This is called "smuggling".
Combat Bonus: The AI cannot move HQ's into range to support op's nor can it prep units for a location. Thus at Hard and VH it will get bonuses to combat that represent HQ's being present and the units prepped. This is referred to as "absent leadership". Think MacArthur.
The list goes on, but I think you can get the drift. There are no cheats, there are simply facets of the AI. All bonuses given to the AI are already present in the game and a human player wisely avails themselves of them at all times. So does the AI ...

Re: “AI” cheating?
Posted: Sun Mar 30, 2025 6:32 pm
by PaxMondo
I should also add that if any of the above is "cheating" to you, then you are a PBEM player. Don't play the AI, get a PBEM game going ASAP. Don't worry about losing the first few games, and be honest about your skill level. You will find opponents. ENJOY!
If the above seems totally reasonable to you, then you can enjoy AI games at your own pace. Be prepared to move onto the Ironman scenarios quite soon as they will represent the challenge for you.
In either case, enjoy the game.

Re: “AI” cheating?
Posted: Mon Apr 07, 2025 1:08 am
by bigred
PaxMondo wrote: Sun Mar 30, 2025 6:32 pm
I should also add that if any of the above is "cheating" to you, then you are a PBEM player. Don't play the AI, get a PBEM game going ASAP. Don't worry about losing the first few games, and be honest about your skill level. You will find opponents. ENJOY!
If the above seems totally reasonable to you, then you can enjoy AI games at your own pace. Be prepared to move onto the Ironman scenarios quite soon as they will represent the challenge for you.
In either case, enjoy the game.
I wonder if anyone is working on a upgrade for AE which incorpoates new AI tech. We might get a super AI badass jap player.
Re: “AI” cheating?
Posted: Mon Apr 07, 2025 1:09 am
by bigred
PaxMondo wrote: Sun Mar 30, 2025 6:32 pm
I should also add that if any of the above is "cheating" to you, then you are a PBEM player. Don't play the AI, get a PBEM game going ASAP. Don't worry about losing the first few games, and be honest about your skill level. You will find opponents. ENJOY!
If the above seems totally reasonable to you, then you can enjoy AI games at your own pace. Be prepared to move onto the Ironman scenarios quite soon as they will represent the challenge for you.
In either case, enjoy the game.
I wonder if anyone is working on a upgrade for AE which incorpoates new AI tech. We might get a super AI badass jap player.
Re: “AI” cheating?
Posted: Wed Apr 09, 2025 3:32 pm
by PaxMondo
bigred wrote: Mon Apr 07, 2025 1:09 am
PaxMondo wrote: Sun Mar 30, 2025 6:32 pm
I should also add that if any of the above is "cheating" to you, then you are a PBEM player. Don't play the AI, get a PBEM game going ASAP. Don't worry about losing the first few games, and be honest about your skill level. You will find opponents. ENJOY!
If the above seems totally reasonable to you, then you can enjoy AI games at your own pace. Be prepared to move onto the Ironman scenarios quite soon as they will represent the challenge for you.
In either case, enjoy the game.
I wonder if anyone is working on a upgrade for AE which incorpoates new AI tech. We might get a super AI badass jap player.
I'm not on the dev team, but I would be stunned if this were to occur. The AI isn't just the scripts, but also all of the other operations that the computer controls as well (logistic movement (supply/resources) for example). They're integrated/intertwined to the extent that to separate them for a new AI player software would be a HUGE task. There has not been a whisper that this is in the works, and the few times the devs did comment about this in the past it was: "it is what it is". So, ...
As an AI player, I would be ecstatic, but I have zero confidence that anything like this is in the works for AE ...
Re: “AI” cheating?
Posted: Wed Apr 09, 2025 4:42 pm
by BBfanboy
+1 to what Pax said. From what I have read, AI must be 'trained' to be able to operate effectively as an opponent. That would require access to hundreds of games and programming what to take as positive moves and what NOT to do based on success of attacks or logistics/industry management. So this pipe dream is like all the other WITP-AE2 dreams out there - too expensive for the potential return on investment.
Re: “AI” cheating?
Posted: Sat Apr 12, 2025 2:02 am
by pbiggar
I went directly to the source on this one. In this case Deepseek. Very interesting that it suggests "an assistant" to help with the next move might be an intermediate step to assist players in completing their turns.
Here is what Deepseek responds to this prompt.
"When will AI be able to function as an opponent in complex war games that many players play PBEM currently. Games like War in East, or War in the Pacific Admirals Edition?"
DEEPSEEK's response (this version was trained up to June 2024):
AI is already making strides in complex war games, but reaching the level where it can fully replace human opponents in highly detailed, strategic-level PBEM (Play-by-Email) games like War in the East or War in the Pacific: Admiral's Edition (WiTP:AE) will likely take a few more years. Here's a breakdown of the challenges and timeline:
Current State (2024)
AI in Wargames Today
AI can handle operational/tactical-level games (e.g., Panzer Corps 2, Command: Modern Operations) reasonably well.
Strategic-level AI (like in Gary Grigsby’s War in the East) is still limited—it follows scripts, lacks deep adaptability, and struggles with long-term planning.
Machine learning (ML) has been applied to games like Chess, Go, and Starcraft II, but these are either abstract or real-time, not turn-based, multi-layered simulations.
Limitations in PBEM-Style Games
These games involve:
Massive decision trees (thousands of units, logistics, fog of war).
Long-term strategy (campaigns lasting hundreds of turns).
Asynchronous play (PBEM requires AI to "think" like a human opponent).
Current AIs (even LLMs like GPT-4) lack the deep reinforcement learning (RL) needed for such complexity.
When Will AI Be Viable?
Timeframe Likely Progress
2025–2027 - Improved RL models (beyond AlphaGo/AlphaStar) applied to wargames.
Copy
| - AI could handle **operational-level** games well (e.g., *Unity of Command 2*).
| - **Strategic-level** AI remains brittle but may assist human players. |
| 2028–2030 | - AI could credibly replace human PBEM opponents in complex games.
| - Requires:
- Massive datasets of human playthroughs.
- Hybrid AI (LLMs for strategy + RL for tactical execution).
- Hardware advances (faster training for large-scale simulations). |
| Beyond 2030 | - AI may surpass human players in all aspects of grand-strategy wargames.
| - Possible dynamic, adaptive opponents that learn from player behavior.
Key Breakthroughs Needed
Better Reinforcement Learning – Current RL struggles with the scale of games like War in the Pacific.
Human-like Strategic Reasoning – LLMs can help, but they need fine-tuning on wargame logic.
Procedural Scenario Adaptation – AI must handle dynamic campaigns, not just pre-scripted behavior.
Conclusion
By 2027–2030, AI will likely be strong enough to replace human PBEM opponents in complex wargames.
Before then, we’ll see AI assistants (e.g., suggesting moves) rather than full opponents.
Games like War in the East will be among the last to get strong AI due to their extreme complexity.
Would you like recommendations for current AI-assisted wargames? Some (like Shadow Empire) already use advanced AI for certain mechanics.