I should probably begin with a confession. I am not a computer nerd. I am just a plain old common nerd, which is an entirely different branch of the species. I know how to drive computers in much the same way I know how to drive a car. I can make them go where I want, most of the time, but I have no real idea what is happening under the bonnet. If something starts clanking, smoking, flashing, or making noises it should not make, I do not reach confidently for a toolbox. I stare at it suspiciously and hope someone cleverer wanders past.

Computers themselves remain, to me, a form of eldritch magic. I press keys, things appear, files vanish, programs sulk, and occasionally the machine decides that what I really wanted was an update at the worst possible moment. Somewhere inside the box there are probably tiny demons, wires, runes, and a committee of disgruntled goblins making decisions I was not consulted about. Artificial intelligence, therefore, sits even deeper in the cave of sorcery. I use it, I prod it, I argue with it, and sometimes I am impressed by it, but I do not pretend to fully understand how it works.
Despite this I am going to have a go anyway!
This post has come about in response to a discussion I had with Soldier of Tin and others about AI and his Postcard Post. As often happens in the hobby, a passing exchange sent me off thinking about something larger than the original conversation. AI is one of those subjects that tends to produce strong reactions, ranging from breathless excitement to pitchforks and flaming torches, with most sensible people standing somewhere in the middle wondering whether it might at least help sort out a campaign roste
That is probably a useful place to start when talking about AI in wargaming. I am not approaching it as a programmer, technician, futurist, or prophet of the silicon age. I am approaching it as a wargamer, rules tinkerer, campaign scribbler, and lifelong collector of too many figures. What interests me is not whether AI is clever in some abstract technical sense, but whether it helps us make better games, write clearer rules, organise campaigns, test ideas, and get more toys onto the table before we are distracted by the next shiny project.
Artificial intelligence has arrived at the wargaming table, whether we invited it or not. Like most new tools in the hobby, it has been greeted with equal parts enthusiasm, suspicion, wild exaggeration, and the occasional cry that civilisation has finally ended. It has not. Dice still roll badly. Players still forget morale tests. Someone still argues about line of sight.
AI is not a magic oracle, but it is already a useful assistant for wargamers, rules writers, campaign organisers, and solo players. As I argue in A Wargamer’s Guide to Writing Rules, modern designers now have access to tools that can help with research, editing, scenario generation, play testing analysis, probability work, and consistency checking, but those tools remain aids rather than replacements for judgement, creativity, historical understanding, and real players at the table.
The first great advantage of AI is speed. It can take a rough idea for a scenario and quickly suggest objectives, terrain, force structures, complications, and victory conditions. For a club night or campaign game, this can be enormously helpful. Instead of staring at a blank page wondering why your “simple raid scenario” has somehow become a three week research project, AI can provide a working draft in minutes. That draft still needs the designer’s eye, but it gets the process moving.

AI is also useful for organising material. Many wargamers are drowning in notes, army lists, half finished rules, campaign maps, and brilliant ideas written on scraps of paper that made perfect sense at the time and now resemble intercepted enemy signals. AI can help sort these into usable structures. It can summarise playtest feedback, identify repeated complaints, compare versions of a rule, and point out where terminology has drifted. If one section says “suppressed,” another says “pinned,” and a third says “shaken,” AI is quite good at asking whether these are meant to be the same thing.
For rules writers, one of the strongest uses is coherency checking. AI can read a draft and look for contradictions, missing procedures, unclear examples, and edge cases. It can ask annoying but useful questions such as: what happens if a unit is both hidden and forced to retreat, who acts first when two reaction rules trigger together, or whether a morale test happens before or after casualties are removed. These are exactly the sorts of problems that cause arguments at the table and errata after publication.
AI can also support probability and play testing. It can help build simple Monte Carlo tests, run through likely outcomes, and expose mechanisms that are much more lethal, random, or predictable than the designer imagined. This does not replace rolling dice with actual players, but it can identify problems before a rule reaches the table. It is better to discover during testing that your elite cavalry are statistically doomed than after three club members have painted forty figures and now want your head.
There are also clear benefits for solo gaming. AI can help generate events, enemy behaviour tables, campaign consequences, rumours, unexpected reinforcements, and narrative hooks. It can make a solo campaign feel less predictable by adding controlled uncertainty. Used carefully, it can become a kind of assistant umpire, not replacing the player’s imagination but giving it something to react against.
The minuses, however, are real.
The most obvious problem is accuracy. AI can be wrong, and worse, it can be wrong confidently. It may invent a unit, misstate a battle, confuse one campaign with another, or produce a quotation that sounds excellent but never existed. This is especially dangerous in historical wargaming, where plausibility can disguise nonsense. If AI tells you that a particular battalion fought at a particular battle, check it. If it gives you an order of battle, check it. If it provides a source, check that the source actually exists. “The computer said so” is not research.
There is also the risk of blandness. AI is very good at producing competent generic material. That is useful, but it can also flatten creativity. A scenario generated entirely by AI may be tidy, balanced, and utterly forgettable. Good wargames usually contain a human spark: a strange historical detail, a designer’s obsession, a club joke, a nasty little tactical dilemma, or a dramatic imbalance that makes the game memorable. AI can assist with this, but it cannot care about the game in the way the designer does.
Another danger is over reliance. If every scenario, rule, army list, and campaign event is generated without judgement, the designer gradually stops designing and starts accepting. That is not using a tool. That is surrendering the craft. The best use of AI is conversational and critical. Ask it for options, then reject most of them. Ask it for weaknesses, then decide which ones matter. Ask it for a draft, then rewrite it until it sounds like you rather than a committee trapped inside a laptop.

AI can also give a false impression of testing. A simulation may tell you that a rule produces acceptable averages across ten thousand runs, but averages do not tell you whether players enjoyed it, understood it, or felt cheated by it. Mathematical balance is not the same as table balance. Human players notice pace, drama, frustration, agency, and fun. A rule can be statistically sound and still be dull. Real playtesting remains essential.
The best way to think about AI in wargaming is simple. It is a very useful assistant and a very poor master. Use it to draft, test, organise, question, summarise, and challenge. Use it to find weak points in rules and to suggest alternatives. Use it to help with the dull jobs that slow projects down. But do not let it replace historical judgement, player observation, design instinct, or the glorious messiness of actual games played by actual people.
AI will not save bad rules. It will not make a dull game exciting by itself. It will not replace the club, the campaign notebook, the argument over the hill, or the story of the militia unit that somehow held off the guards cavalry for three turns.
What it can do is help us write, test, and present better games.
And if it can help us get one more half finished rules project onto the table before we are distracted by another period, another army, or another box of unpainted figures, then it has probably earned its place beside the dice, the ruler, and the cold cup of tea.
I have used ChatGpt for the illustrations in this post.










































