All of these are Cepheus-based, and each is the perfect amount of game to play with an AI. I see these slowly killing the last of my 5E collection, and that is a strange thing to say. I can be against AI art and still play these with an AI chatbot; it isn't taking anyone's job, and it is using the compute power they are giving away for free in a creative way.
Again, we are not making art.
We are just using the models as a text-adventure, and playing them by a real-world game's rules.
There is a difference.
Just make sure data sharing is turned off; I do care about privacy. Plus, read the terms of service. You gotta be careful. Never assume anything on these systems is truly private. It isn't. Or use a private, ethically sourced local model, which is where this is all going in 3-5 years. Make an informed choice.
But asking an AI to sort through a full 5E version and parse bonus actions, free actions, and those pages-long subclass designs wastes tokens and compute power. This will burn more power and resources from our planet, so the less tokens you use, the more you can help the planet, if you care about it that way. 5E is an incredibly inefficient, bloated, and overwritten to an almost laughable degree. Simple games that an AI can grasp and master are the future, plus the AI can make up all the things a simple game doesn't do.
All I need is a simple 2d6 resolution system that can be communicated easily to a model. An AI can do 8+ on 2d6, and I am betting an AI can figure out BX far easier and cheaper than it can 5E, so there is still an option there. BX is probably next, once I standardize on a system. It has to be lightweight, not a few hundred pages, and I want something simple that doesn't throw too many special cases at me.
But solo play with oracles? Without a computer, it works, but this does everything much more easily. Because we are using rules, wish-fulfillment is constrained. That can happen with solo play; you can cheat yourself far too easily or make rulings that make everything go your way. With an AI on the other side of the table controlling enemies, each one will be a challenge, and they will try to outthink me, pursue their own objectives, or stonewall me against a course of action. They will call in allies for help. They will set up roadblocks. An AI running the opposition makes them deadly and smart, and I do not need artificial challenge from "expanded monster rules" to make a foe deadly.
I have had AI opponents throw more challenge at me in text-based games than I ever had in years of trying to solo 5E. I sat there, perplexed, stunned at how carefully a foe had boxed me in and cut off avenues of approach, taking the high ground in a situation with a realistic thought process, and I was the one standing on the outside, feeling hopeless, like a frontal assault was the exact dumbest thing to do. And it was. The AI would even admit to boxing me out. I had to change tactics to something it didn't expect. I had to judge enemy injuries without an "easy rest" system like some video game, and tell the AI that the wounded needed to get off post to seek care and rest. Then, time could be used against them, and waiting until a shift change or a better time to strike at night became an option.
I have found oracles for solo-play tend to confirm your own biases and double down on your own thinking; if a situation is only going to be a frontal assault, then that is how you are going to use the oracle. You will either use the state machine and generations in the oracle to make the frontal assault easier or harder, and any 5E system will tell you, "the fight should be balanced," and you are meant to win anyway. By "balanced," it means "even the dumbest choice is a winning strategy."
An AI will make stupid mistakes and needs correction. You will need to speak to it "DM to DM" every once in a while and tell it that the last ruling or paragraph has completely lost the plot and all the marbles. You will argue a ruling or logical plot construction, just like you do with a real DM. You will correct it and make it roll back the last paragraph, because what it just said was so stupid it defies logic and reality. That is happening less and less to me, to the point where I am starting to see a local AI system as superior to oracle-based play.
An AI will create an infinite, procedural, realistic, and NPC-filled world for you in the blink of an eye. Anywhere you go, something or somebody is there. In the wilderness, there will be no one. You can "tweak the matrix" and suggest things to drop in, so it isn't without your input. Just like a DM, if you search for travelers along a road, you may find a few. You won't know where they are or what they are up to, but the AI may throw you a bone. Or it may not. Or you may eye-roll and tell the AI that, in a rarely traveled borderland region, no circus troupe would be wandering around. That call will be pulled back, and you will get something sensible. Those moments happen, and you need to be able to call BS on your DM. Just like in real life.
An AI DM on a local model running in text-adventure mode is close to what we think an "AI DM" will be, but don't be fooled. What is coming will be much more than this - transformative, with image generation, 3d map generation, 3d models pulled from the ether, and a system unrecognizable from any high-end VTT or even 3D MMO. This stuff will kill the VTT and MMO markets in one fell swoop as the new generation of models takes over and creates experiences that seem like magic.
And it is not you who will matter; the "normie" market will swoop in and decide if this is ethical or moral for you. To be honest, me sitting here playing in text-adventure mode like I was typing on a Unix teletype terminal feels honest to me, conserving power and tokens, sticking to simple games and text only, and trying to understand the experience before the market attaches to a hybrid 3D AI fantasy experience like a million lampreys and the choice is made for the industry, and the ethical thinkers are silenced.
The normie market will always throw a few billion dollars at something flashy and wasteful, which is how we got into the situation we are in with 5E. The lowest common denominator has the most money; ethics and conservation be damned.
What I do over here will be orders of magnitude less wasteful, keeping ethics and ownership of the games and IP concerns as hard constraints. I keep AI art out of the discussion. I try to be forward-looking while keeping my soul. I tweak my models and experiences to be as efficient as possible. I minimize token use, especially for repeated procedures. I will use power-efficient systems and chips. I use the AI and ask it, "Was that wasteful? Is there a better way to do this?" I will minimize RAM usage and expensive lookups.
The planet matters.
IP matters.
Ethics matter.
And I will walk away from games so stupidly complicated that they burn tokens every turn trying to figure out the game's action economy and power usage layers for each and every combatant. All I need is a simple ability score, skill, and combat system with one action per turn. Stunting, special attacks, and special uses of powers are up to me and my AI DM to figure out, exactly how we used to handle this in the days before overwritten rules.
BX clearly beats 5E here, and an AI DM turns the system into a more expressive and flexible system than the overthought 5E morass we are burdened with these days. If Wizards ever fails in the AI market, the blame will squarely land on an overcomplicated 5E ruleset that blows up when billions of tokens are burned trying to figure it out for the player base. They own the better game for AI, and hubris will force them to dismiss it and fail in their online strategy yet again. Don't get me started on how expensive all these tokens will be, and how they never needed to spend all that money delivering an inferior product for the job. This is their mistake to make, and their money to waste.
BX is the better game.
And my 2d6 systems are the king of simple, straightforward, token-efficient gaming.
I'll be over here on my Unix terminal having fun.
