By A. Umbral
Cthonic Creations
OSR
Level 1
Beneath the shadowed walls of Crow’s Keep, treason festers in whispered secrets and quiet deals. The war-weary King Uldred fights to hold his crumbling throne, while unseen forces conspire in the dark corners of noble halls. The city’s watchful Reeve has sent word to a handful of expendable operatives—those desperate enough to gamble their fate on a mission veiled in secrecy. Their charge? To infiltrate an uncharted cave deep in the foothills, where a bandit faction has taken root. But steel alone will not be enough. Beneath the carved ruins of forgotten empires, something far older stirs. A hidden temple lies buried under the earth, its walls heavy with serpentine scripture, its chambers thick with the weight of ancient curses. Lady Rinwolde’s network has already reached this place, her spies clawing at long-lost relics of unfathomable power. Blood will spill in the darkness. Trust will be tested in the fires of ambition. And in the ruins where mortal and monster meet, the truth is as sharp as the blades that gleam in the dying torchlight. Will you uncover the mystery before Crow’s Keep collapses into war? Or will you vanish beneath the earth, another forgotten name swallowed by history?
This thirty page adventure uses about four pages to describe about 45 rooms across two levels of a bandit lair/snakeman temple. Abstracted and minimalistic in the dungeon, while trying its best to hit all of the marks of a good adventure. I am generally left confused on the choices made for an adventure outline.
Communicating the vibe of something is hard. I generally push in to hyperbole, trusting that my intelligent, good looking, and humble readers can follow along. In this case, what if you had a Vampire Queen dungeon of 45 rooms over a few pages, that really aggressive minimalism that showed up there. And, then, as preamble I stuck in a modern intro and hex crawl and then in the appendix included a massive rumor table and monster stats and lore and so on. There would be this massive disconnect, right, between the amount of detail that The Main Event has vs the supporting information. It’s not that the dungeon MUST be the main event; it could be a village social thing or it could be a hex crawl thing, with the goal being a small dungeon or some such with The Thing in it you want to get. In these cases it would make more sense for more effort to be spent on the hex crawl or the social village elements or some such. But, if the dungeoncrawl IS the adventure then I must point out the obvious disconnect. COULD you write a five page dungeon that is great inside of thirty pages? Sure. Does this? No.
Ok, you’re level ones and the default hook is that the local reeve is sending you to check out some bandit caves. Seems you’re convicts and you get a pardon if you do well, whatever that is. You’re sent to spy, learn information, and so on. Absolutely nothing in the adventure is going to help you do that, that’s unsupported in every way, but that’s the pretext. You’re taken by a ferryman (with some decent read-aloud, all in italics, alas, but nicely done) across the water to your start point. You’ve got three days of iron rations and he’s coming back for you in five days, no more no less, and not waiting around for you. You’ve got a two day “hex crawl” in front of you till you reach the bandit lair. This is all looking a little rough for level ones … a strict timeline doesn’t really mesh well with the hit and run away vibe of squishy characters. It’s a very structured “hex crawl” in that the DM is essentially rolling for wanderers at the appropriate time but everything else is very controlled. Roll on the weather table. Heal a HP if the night was chill, you enter a mountain hex, etc. The wandering monster table for the dungeon is also a bit more than I expected. “1 Escaped Prisoner – caught by bandits. Could be adventurers, possibly allies.” or “Standing Water in Passage – water pit, 5 feet deep. Slows characters. “ These are both ok things. I’m in a pretty pleasant mood at this point and looking forward to the dungeon.
Then comes the dungeon map. This is a half page thing, full color, mage in Dungeongrapher or some such. Lots of textures and tables and shit on the map. It’s a disaster. Too small, too much detail and overlapping textures. There’s no real complexity to the map, but, also, it’s barely legible, which is a problem.
Next up comes a summary of the various rooms in the dungeon. This is something I sometimes come across. I understand the goal, but I think it seldom works out the way the designer wants. In this case, it’s presented in two column table format. The first table column has a room name and maybe a couple of details why the second table column has a few notes about the room. “Entrance Tunnel” and then “Narrow stone passage littered with old bones of animals. A makeshift barricade with a single guard.” So, sure, that’s fine. Sometimes the first column has a few more details, things that might be obvious to be seen and so on.
Oh.
Wait.
That’s not a summary.
That’s the actual dungeon.
Mind you, room two, which I’m about to quote, is INSIDE a cave: “Guard Watchpost – each tower has a Bandit Guard “ That’s the first column. Then column two of the table says “Elevated overlook where scouts track movement. Two small wooden towers.” Repetition. Minimalism. Abstraction. Sometimes monsters (bandits) show up in column one. Sometimes in column two. There are never more than a sentence or so of words. “12 Hall of Murals “ and “Bandits have partially uncovered ancient serpentfolk murals-some have begun whispering in their sleep. “ These are ideas, not encounters. You’re stabbing folks. There’s no infiltration here, there are not supporting notes for that of any type. Stab Stab Stab! Sure that’s fine, I guess. Sometimes.
How about an EXTENSIVE rumor table! “Whispers from the Past” – “Superstition” “True (Cave Wraiths whisper in lost languages, and some bandits are driven to madness)” Uh. What? What’s the purpose of the rumor title? Am I missing something? There’s two fucking pages of these. I don’t know, forty, fifty of them? Like I said, a minimalism dungeon supported by everything you would want in a lot of detail. But, in a weird fucking way. I’m not sure I know how to use that rumor table. It’s like the heading title is supposed to have more information or something? But it doesn’t? I don’t know.
One of the rooms has a trap. “Door Locked with a simple trap.” That’s it. You want to know what the trap is? Do the work yourself I guess. No order of battle. No infiltration notes. No real tricks or traps, given a definition of what a trick or trap would generally be agreed to.
Unless you REALLY know what you’re doing, pay attention to the main adventure. That’s where your effort should go. I’m at a loss as to how that can be a mystery, but, there you go.
This is $10 at DriveThru. The preview is seven pages. You get to see the intro, hex crawl, and the first seven rooms. No, that’s not a summary. That’s roughly 20% of the dungeons rooms. Good preview?
https://www.drivethrurpg.com/en/product/530730/cursed-blood-and-cold-steel?1892600
After reading your review, I thought this just sounded, once again, like AI. So, I checked the preview. Yep. It’s copy-and-paste ChatGPT. The tables, bullet-points, bolding, the names it generates—all of it—is what ChatGPT looks like when you copy-and-paste it into Google Docs and don’t even bother to edit it.
(He probably did guide and edit it some, though. Like, maybe some or even many of the ideas were his, and he probably didn’t just generate this all at once and take the first thing it spit out. This took some human effort, at least.)
After the whole Reality Unbound: Reclamation (*spits*) comment section absurdity, I hate to say anything, but . . .
Bryce, it’s not that you’re naive. It’s not “fooling” you into thinking it’s human. (It does that to literally everyone.) You’re not just “bad” at “detecting” it. I don’t have some sort of sixth sense or whatever. And, I’m not even busting your chops, man.
It’s simple: the more you use it, the easier it becomes to recognize.
No one can “detect” it. You can only “recognize” it when you know, from experience, what it produces. You say you’ve used it, but you’ve not used it nearly enough if those bullet points and that table didn’t make you think, “This is ChatGPT!”
One thing people have talked about is that ChatGPT uses excessive em dashes. Usually at least one per paragraph. You can tell the author wrote the backstory. It uses en dashes with spaces around them. “How this ties into the adventure” uses way too many em dashes with no spaces. That’s how an editor would do it but this guy didn’t hire an editor. And an editor would not use them in almost every sentence.
I had to tick a box to “prove I’m a human” before I could download the preview from DT. The irony.
This is so interesting. You’re right; I’ve clearly not used it enough to recognize it.
I wonder, though, why did the designer use it?
There is no fame in writing an adventure. There is absolutely no money to be made. This leaves, yes, the simple joy in creation? Then automating that away gives you nothing, as the creator? Right, there is no job to get, class to pass, BS to ignore, art to fill in or thing to do cheaper. If you wanted to READ a novel I can understand having AI write you one to read. But it you wanted to create a novel … why automate it? If it’s not your job, and you’re not posturing … why? “ChatGPT, watch every episode of every season of Andor and tell me what to think about it.” I’m not being facetious, I don’t understand this.
There isn’t a lot of money to be made in adventure writing, sure, but I wouldn’t say there’s *no money* in it, especially if the “author” is charging $10 a pop. This strikes me as one of those passive income things.
Not too long ago, YouTube was flooded with videos for how to make passive income using AI to generate kid’s books, posters, and coloring books to be sold digitally. Making one of them won’t make you more than a few dozen sales, but dozens of them could net you a nice little cash flow – a cash flow that requires next to no invested effort or expense. I suspect this adventure may be the tip of one of those icebergs.
Practically, how old do you need to be to create and sell an AI moduled via drivethru or itch?
Maybe the “authors” are kiddies?
Seeing as there’s little to no quality control or preventative measures to using AI or opening an online sales account (other than having your own bank account), I would say age is not a real barrier to any of it.
There was a recent controversy, revealed at DF, where an author’s use of AI sucked in *an entire work* posted at DF’s Workshop, spat it back out nearly verbatim -without attribution- and the “author” published it as his own workthrough DMsGuild and DTRPG, where it became a “Silver Seller.”
https://www.dragonsfoot.org/forums/viewtopic.php?t=94228
After the actual author complained, it was pulled down by both outlets. I don’t know if the plagarizer shared some/all of the sale proceeds. However, it’s scary to see how that happened. If you’ve self-published or even posted gaming material on a gaming messageboard, it may end up being nabbed by AI.
Years and years ago, someone tipped me off that an eBay seller was printing out my modules published at DF (which are free) and was selling them. I thought that was bananas; I would be apoplectic to see someone publishing my work as their own through lazy use of AI.
And, I should point out, the plagarizer responded in the thread – using AI-generated responses.
What a world, what a world.
It allows people to create professional-looking products with minimal effort. Writing a module, getting a decent map, lining up human artists – these things take effort vs. a few clicks.
Once I realised that I started writing better adventures. Sure, people like DM Dave swim in cash I guess. But let’s take me as an example.
On drivethru,
I have about 10 5e adventures since my patreon years. They generate next to nothing.
I have 2 PWYW OSE adventures from my “getting comfortable writing for OSE days”
And I have 2 newer OSE products that received the best and no regrets from you.
Deivetgru sends me a reveit every 2-3 months for like 150-300 dollars, depending on how fresh my newest product is. I see it, I smile, I think cool. Gas money. And move on. There is no real money to be made here, and all we need to strive for is to create a memorable product. Because we love the creative process. Not because we want to get rich.
Having said that, I am a defender of AI under certain circumstances. As a Google answerer that is directly answering a specific question.
Example. My new adventure is based in byzantine Greece. Semi historical. I asked chat gpt, which rooms would a typical monastery of that era have? It gave me a list. I picked the ones that made sense for my vision… fleshed them out myself. I could have googled this, visit 10 websites with a million ads on each, where the information is hidden somewhere at the bottom because they get paid by the word.
AI isn’t necessarily a demon; it can be used for good or evil.
Using it to research a topic and get some quick answers to questions is good.
Using it to write an entire adventure with little modification so that you can sell it to suckers is evil.
>I’m not being facetious, I don’t understand this.
I promise: you would better understand if you use it—as I’ve suggested in each of your comment sections for a long time now—to write a GOOD (eight-page?) adventure.
First of all, your premise is off.
Without your heavy involvement, ChatGPT cannot write a GOOD adventure. You will have to be the primary author if it’s to be a GOOD adventure.
AI might have fooled you (like everyone else) into thinking it’s human—but it has never fooled you into thinking it could write a GOOD adventure, has it? Nope! You’ve ripped them all to shreds!
AI is a just tool for authors to use. Just like spell-check. It will make a text better, but it won’t make it good.
Log in to ChatGPT with your Google account for free and type something like this:
“I want you to help me write an eight-page adventure using the 1981 edition of the Dungeons & Dragons Basic and Expert Sets. How shall we begin?”
Keep an open mind. It’s not always going to do a good job, but you will be able to work with what it gives you. Plus, you teach it as you go. Tell it to use evocative but terse sentences. Tell it to avoid purple prose. Tell it to put the important stuff up front in a description. Talk to it like you would a human student writing an adventure. Give it the same advice you give everyone on here.
When you’re finished—and that will take a lot of time—you’ll have an adventure AND a far better ability to recognize what it outputs!
The Drivthru page say “Creation Method: Contains AI-Generated Content”. I’ve never seen that before. But that could easily be me.
I enjoy my players having fun with my nonsense creations, so I am at a loss with this. But to be fair, I’m confused by so many people running 5e corridor shooter adventures and thinking they’re what it’s all about.
Speaking with ignorance and not a little dread; I wonder what AI would produce if trained on good adventures?
Cover art is procedural, too. Warrior’s hand is backwards, repeated beats in the wall art.
Clearly the warrior is a rakshasa in disguise.