Use AI for Routine TTRPG Prep, Not for Creative Work

The Halls of Arden Vul by Richard Barton is a truly massive megadungeon. It’s a love letter to AD&D and the classic megadungeons of the 1980s.

The good part of the classic feel is obvious. Listening to the 3d6 DTL podcast’s long-running Arden Vul campaign, it’s impressive to see the density of factions, the depth of the dungeon’s history, and the opportunities for interactivity. I would love to run it someday. 

The bad part comes down to formatting and at-the-table ease of use. I wouldn’t say it doesn't do anything to incorporate the advances TTRPG usability over the last 40 years; it includes expansive overviews and aggressive cross-referencing, an improvement from the “figure it out yourself” opacity of a lot of ‘80s D&D products. 

But it doesn’t go nearly far enough. The text is dense, written for casual reading rather than for quick reference. Ancient history is intermingled with recent events. There’s little use of bullets, bolding, or sidebars. Even little things reflect a retrograde formatting mentality; listing damage as “2-7” rather than “1d6+1” is a small detail, but it is a pointlessly archaic Gygaxism that can trip people up; particularly new DMs, or those who don’t easily intuit die increments from numerical ranges. Compare this to a modern product, like Necrotic Gnome’s OSE releases, with their signature house style, carefully calibrated for use at the table. Arden Vul is old school not just in play culture, but also in formatting and presentation.

Don’t take my word for it. Read Bryce’s review on Tenfootpole. He obviously has high praise for the amount of evocative detail and interactivity on display in Arden Vul. But his reviews also place a premium on ease of use at the table, and many of his criticisms of Arden Vul land here. See the “Jhentis the Ghoul” section of the linked review for an apt summary of the problem.

Browsing my own copy of Arden Vul, I have to agree with him. It is frustrating that such a massive, fascinating work should be held back by mere formatting and organization issues. Implicitly, the DM running this is expected to make up the difference. As Bryce puts it in the review: “Highlighters out! Actually, better buy a gross of them, you’ve got 1122 pages to read, absorb, and highlight.”

But it’s 2025. We aren’t limited to highlighters, right? Can’t we use… technology? Can we get the AI to do it for us?

On the Use and Misuse of Generative AI in TTRPGs

Generative AI is a hot topic in many spaces, including indie creative scenes like TTRPGs. While some indie and OSR creators have dabbled in using generative AI, there's a lot of ire directed at these tools as well, particularly from artists who object to Midjourney and similar image-generation software.

I’m sympathetic to concerns about AI content replacing human-made content. In the past, I used AI images on this blog, just to add some color to my posts. But recently, I’ve been losing interest in those image-generation tools. While my use of them did not displace any work that would have been commissioned to a human artist, I’ve nonetheless found that as AI-generated images have gotten “better,” they’ve lost some of the weirdness they had just a few years ago, and are turning into professional-but-bland tools, better suited for workplace PowerPoint presentations than TTRPG content. There’s much more to say on this topic, but that’s a whole post of its own, for another time.

AI image generation in action. Technically, yes, this delivers on the prompt request of "robot rearranging things in a dungeon room." But everything in it is either sharp-but-derivative (the very generic robot); blurred to the point of abstraction (the figures, the chest (?) on the left); or failing on a basic level to model a coherent 3D space (the archway, the candles, the stairs, the robot's lower body). Image generation like this is not a good replacement for human-made creative work. 

But this skepticism of generative AI as a replacement for creative work shouldn’t blind us to the usefulness of AI as an assistant for routine tasks. AI tools in their current form are best at handling the least-interesting and dullest parts of the process, freeing up creators to focus on the really creative stuff. And one of the places AI can help the most is in session prep.

Next Week: Can AI Turn an Arden Vul Room Key Into Table-Ready Notes?

TTRPGs Can Make Even Tic-Tac-Toe Interesting

I’ve written previously about how puzzles should be incorporated in RPGs and why tension and escalation makes them interesting. Let’s approach it from another angle and see how far a dead-simple system can go in powering a TTRPG scenario. 

Take, for example, tic-tac-toe*. Tic-tac-toe is a game many children learn to play. But they do not play it for long. They learn the rules, optimize their play, reach an equilibrium where no one can win, and then stop playing the game. You would not ask anyone (besides a young child who has not yet learned the rules) to play tic-tac-toe with you. It would not be a fun experience for either player. In a way, tic-tac-toe is like a very simple lesson in how to enjoy games, and by extension, how to design them.

Now think about Hollywood Squares**, the game show that has been on and off television for nearly 60 years (revived once again this year, Wikipedia tells me). It's a quiz show where players win by connecting a line on a three-by-three grid of celebrity panelists. 

Nobody is watching Hollywood Squares for that hot tic-tac-toe action. But everyone watching brings their understanding of tic-tac-toe to the experience. It provides a framework for the real entertainment. Tic-tac-toe is not in itself fun, but the show wouldn’t have lasted as long as it has without the structure tic-tac-toe provides.

Apply this idea to TTRPGs. You could play a “game” where both players roll a d20, and whoever gets the highest result wins, and there are no other mechanics. This would obviously be a boring game. But a lot of action in a TTRPG session basically boils down to “two people roll, and the better result prevails.” It’s interesting because we care about the fictional events the roll is resolving.

So a mechanic can be too flimsy to stand on its own, but can still serve as a tool in a larger game.

Imagine the following scenario (set aside for a moment the artificial premise and how or why this puzzle would exist in-fiction; this is an example only). A room is divided into nine sectors. Two opposing sides are in conflict to control it – say the adventuring party and an antagonistic faction. Creating a tic-tac-toe streak “wins,” allowing one side to prevail over the other. What constitutes marking a square is up to the DM; maybe just physically controlling a space, maybe overcoming an opponent in one-on-one battle, maybe something else. 

The players will quickly recognize that they’re playing tic-tac-toe, but the scenario is obviously more interesting than “regular” tic-tac-toe. For one, as described above, tic-tac-toe is facilitating something larger than itself, rather than carrying the entire weight of being fun through its rules and mechanics alone. And two, the open-ended nature of TTRPGs means that tic-tac-toe is only a loose framework, not a rigid structure. 

Can a marked square be returned to its blank state, or flipped by another team? Can PCs mark more than one square at a time, or otherwise break the turn order? Can they stop the opposition from marking a square? How can their characters’ abilities and equipment and knowledge be adapted to this unexpected scenario? 

Remember that many games and game systems can be combined in unexpected ways to create something that is more than the sum of its parts.

*AKA Noughts and Crosses

**AKA Celebrity Squares

Worldbuilding Implications of the Megadungeon as Literal Underworld

Last Week: Death in Depth Within the Mythic Underworld

Last week we explored how to make the mythic underworld a literal one as well. This adds a risk/reward mechanism without defanging death, and rescuing an ally from death provides a reason to delve deeper into the dungeon than the party otherwise would.

Does this concept mean that the dungeon is loaded with ghosts? Well, yes, but not the undead wraiths and banshees the PCs are accustomed to. The typical shade is an invisible, barely perceptible presence. 

But they can subtly influence the dungeon. A door slams shut when no monsters are nearby? A shade, agitated to a brief moment of corporality. A new torch goes out in a dungeon hall without so much as a gentle breeze? Snuffed by a shade.

But otherwise, shades are unseen and undetectable, even by magic that finds the invisible and the undead. The lost soul the PCs search for is the exception. Because of their personal connection to the dead adventurer, the PCs can spot a soul they know, once they are close.

"But my gear! But my loot!"

What if a PC dies, and the die roll indicates a result deeper than the deepest level of the dungeon? That means, at minimum, that the soul cannot be rescued until the DM decides to expand the dungeon. Indeed, in a meta sense, the deaths of powerful adventurers may be the genesis of many deep layers of megadungeons. Perhaps the newly created levels reflect the soulscape of the doomed PC that was banished there. The DM should create new levels at their leisure, and notify the players when a rescue is possible.

This works best in megadungeons where the dungeon becomes strange and inscrutable in its lower levels. Not just “very old tombs” or “entrance to the underdark” strange. Strange like ancient mythological beliefs that deep caves were literally the access point to the afterlife. The idea that the underworld is a physical space connected to the normal world, where one can go, and – at great risk – attempt to rescue someone is, of course, part of Greek myth, as well as various other mythologies. This idea could work best in a setting that folds the planes into the “real” world, rather than treating them as alternate dimensions or foreign planets, the way some of D&D’s settings have in the past.

Death in Depth Within the Mythic Underworld

One of the signature motifs of the OSR genre is that the dungeon is the mythic underworld. What if it is also the literal underworld?

If a PC dies within the dungeon, their soul is banished to the underworld… which is to say, a deeper part of the dungeon. How deep? Take the character’s current level, round down to the closest standard die size used in the game, and roll that die. That’s how many levels down they go from where they died.

For example, say a fifth-level character dies on the third level of the dungeon. The player rounds down from five to four. They roll a d4 and the result is a 3. The current level plus three means their soul is now trapped on the sixth level of the dungeon.

The soul might descend straight down, or they may be sent to some random part of the new level. The character’s soul cannot do anything about their predicament on their own. They are a shade, lacking the physicality and ego to alter their situation. They will drift endlessly down here… unless their companions decide to come rescue them.

"Babe please come back you were only 30 XP away from leveling up"

Complicating Rescue 

Exploring the dungeon is a dangerous idea. Attempting to save a soul trapped in the dungeon is much riskier. The additional danger can take a few forms.

Bargain

The rescue party has to make a deal with someone to take the rescued soul back up to the surface. The doomed soul is literally bound by high magic (ninth-level equivalent or greater) to the dungeon floor they’re on. But there’s someone on this dungeon level who can make an exception. This might be a literal lord of the afterlife, gatekeeper, or psychopomp. Or it may be some ordinary dungeon faction or monster that has been gifted this authority by the dungeon’s gestalt consciousness. Either way, this is no ordinary jailbreak. The PCs should be expected to do something difficult or give up something precious to bargain for the lost soul. 

Fade

The lost soul is fading. Shades only last so long before their consciousness dilutes into the ambient fabric of the underworld. If you want to rescue them, you’re on the clock. It is at the DM’s discretion how much time the PCs have, but it should be known to them (or discoverable), and the more powerful the lost soul, the shorter the clock. 

Exchange

The soul can be returned to the surface, but another soul must take its place. It must be a willing soul, not some innocent or random dungeon denizen kidnapped from an upper level (or at least, it can’t be without serious consequences for this crime, which is not just an ethical and moral issue, but also a violation of the rules of the afterlife). It must be a member of the party, or someone they care deeply about, and the sacrificed soul can’t later be rescued in turn; they’re gone forever.

Geas

The rescued soul can leave, but only because Death has found a greater purpose for them. Perhaps some far greater transgressor is out there in the world – a lich or another immortal whose soul, so long withheld from the reaper’s rightful grasp, has become a rich prize. The rescued soul is bound to pursue whatever quest Death gives them. As with other consequences, this is high magic, and is (almost) impossible to subvert or cheat.

Next Week: Worldbuilding Implications of the Megadungeon as Literal Underworld