AI & TTRPG's

[Late Bloomer]

What are you thoughts on AI and how it relates, can be used, its place, in TTRPG's? Do you have experience using AI with your favoured roleplaying system?


Here is an article that I found interesting.

https://www.theregister.com/2023/08/19/chatgpt_dnd_dm/

 

[Bester]

What are you thoughts on AI and how it relates, can be used, its place, in TTRPG's?

Currently, not much use. Also, Claude writes more imaginative stuff than ChatGPT if you're going to try it yourself. A writer I'm following tried using Claude and has left a couple of blog entries regarding his experience:
https://alexanderwales.com/adventures-in-ai-text-generation-pt-1-of/
https://alexanderwales.com/adventures-in-ai-text-generation-pt-2-of/

As for the future, it's a matter of how fast you think it'll develop its writing capabilities.

Remember how Dall-e was drawing 200*200 pictures vaguely reminding you of your prompts, and compare it to what it can do today - draw beautiful stuff in a style that you specify, e.g. the famous styles like SamDoesArt.

I don't want to go into the specifics, but I'm watching it closely and I think in 2025 ChatGPT will be able to simulate your favorite authors. This year, they'll allow fine-tuning, which is what's necessary for this. So in 2025, AI will become very relevant for all forms of writing, for TTRPGs as well. But not yet.

 

[Tyranicon]

What are you thoughts on AI and how it relates, can be used, its place, in TTRPG's? Do you have experience using AI with your favoured roleplaying system?


Here is an article that I found interesting.

https://www.theregister.com/2023/08/19/chatgpt_dnd_dm/

I'm custom-building one to act as a type of DM for a cRPG.

But if you're using chatgpt or something to run a campaign, then yeah it'll probably be best to set it to an assistant role like in the article. Not quite yet ready to take on a full DM role unless you want some really funky but somehow still bland storytelling.

I just did some testing, and apparently ChatGPT is very iffy with combat and violence in general. Lame.

 

[Bester]

I'm custom-building one to act as a type of DM for a cRPG.

Currently Steam allows no AI-related stuff, including games that integrate chatgpt, so good luck publishing that.

 

[Tyranicon]

I'm custom-building one to act as a type of DM for a cRPG.

Currently Steam allows no AI-related stuff, including games that integrate chatgpt, so good luck publishing that.

Custom-built means not chatgpt or any kind of existing LLM. It's built from scratch by me and through scripting rather than code.

 

[0wca]

TTRPGs were meant to be played with friends and one of the major selling points of them is that you have a human behind the helm and not a computer.

If you replace the human GM with a computer you're essentially playing a video game without all the graphical bells and whistles.

If you follow down this trajectory, D&Done will start to look like a good idea and the hobby will basically die because it's the GMs who buy the games and therefore keep the hobby alive.

The only benefit that I see from AI in this hobby is if it is used as a tool to simplify stuff like GM notes, session prep and potentially combat calculations.

 

[ind33d]

What are you thoughts on AI and how it relates, can be used, its place, in TTRPG's? Do you have experience using AI with your favoured roleplaying system?


Here is an article that I found interesting.

https://www.theregister.com/2023/08/19/chatgpt_dnd_dm/

should only be used in sci-fi games like traveller or 2020 or for droid npcs in star wars d6

 

[Akachi]

Only use right now is for generating character and map art. It's hard to think of much use other than art for AI in TTRPGs. Even if you want something advanced and computerised that can do the math for you, all you need is simple scripting for that, not AI, since FoundryVTT can automate all of that already with JavaScript. You can use AI to help with writing and ideas if you're DMing, but besides that taking the fun out of it, you will have better results and more importantly improve your own skills by doing it yourself. Doing it yourself is even more important since probably more and more people will rely on AI to do those things for them with time, meaning there will be a lot of people with zero skills, creativity, and no dedication.

In the future there'll no doubt be AI DMs, but even if they get very good, unless you're an autist, they'll be a poor substitute for the social experience of playing TTRPGs with friends (short of strong AI and buying sentient robot D&D TTRPG slaves you can keep in your basement). Future of AI in RPGs is more important for video games, not TTRPGs.

 

[catfood]

If you follow down this trajectory, D&Done will start to look like a good idea and the hobby will basically die because it's the GMs who buy the games and therefore keep the hobby alive.

That's exactly where Hasbro wants to take it. Their end goal is to move the hobby exclusively online where it can be controlled to the smallest detail, together with a healthy dose of micro transactions too of course.

 

[JamesDixon]

If you follow down this trajectory, D&Done will start to look like a good idea and the hobby will basically die because it's the GMs who buy the games and therefore keep the hobby alive.

That's exactly where Hasbro wants to take it. Their end goal is to move the hobby exclusively online where it can be controlled to the smallest detail, together with a healthy dose of micro transactions too of course.

Along with a subscription fee too boot.

 

[Bruuce]

Someone made a Foundry VTT module that adds chatbot NPCs to your game.

 

[0wca]

Only use right now is for generating character and map art. It's hard to think of much use other than art for AI in TTRPGs. Even if you want something advanced and computerised that can do the math for you, all you need is simple scripting for that, not AI, since FoundryVTT can automate all of that already with JavaScript. You can use AI to help with writing and ideas if you're DMing, but besides that taking the fun out of it, you will have better results and more importantly improve your own skills by doing it yourself. Doing it yourself is even more important since probably more and more people will rely on AI to do those things for them with time, meaning there will be a lot of people with zero skills, creativity, and no dedication.

In the future there'll no doubt be AI DMs, but even if they get very good, unless you're an autist, they'll be a poor substitute for the social experience of playing TTRPGs with friends (short of strong AI and buying sentient robot D&D TTRPG slaves you can keep in your basement). Future of AI in RPGs is more important for video games, not TTRPGs.

Honestly, having the AI come up with ideas for you is asinine to me. You'll never grow as a GM if you do that and every game that relies on it will eventually devolve into a homogenous soup of the same overhashed ideas.

 

[udm]

I use AI to help me come up with ideas from time to time, but they are very homogeneous and after about 10 or so ideas, you start to see repetitions. What I do in combination with AI is to stock up an entire library of modules that I can study and reference for developing ideas. I run about 5-8 games a month for 8 different campaigns, so I take my inspiration where I can get it—cRPGs, PnP modules, even non-RPGs like Thief and HOMM.

But yeah relying purely on AI is not a good option at all. The way it's programmed is just too limited. These fuckers write essays that can be easily detected by the naked eye because they have a strong tendency to paraphrase their words over and over again.

If you follow down this trajectory, D&Done will start to look like a good idea and the hobby will basically die because it's the GMs who buy the games and therefore keep the hobby alive.

That's exactly where Hasbro wants to take it. Their end goal is to move the hobby exclusively online where it can be controlled to the smallest detail, together with a healthy dose of micro transactions too of course.

Along with a subscription fee too boot.

Actually I hope they really go through with this plan. With luck, maybe it will show the world how retarded it is after WOTC has sunk millions into this pipe dream.

 

[Serus]

Only use right now is for generating character and map art. It's hard to think of much use other than art for AI in TTRPGs. Even if you want something advanced and computerised that can do the math for you, all you need is simple scripting for that, not AI, since FoundryVTT can automate all of that already with JavaScript. You can use AI to help with writing and ideas if you're DMing, but besides that taking the fun out of it, you will have better results and more importantly improve your own skills by doing it yourself. Doing it yourself is even more important since probably more and more people will rely on AI to do those things for them with time, meaning there will be a lot of people with zero skills, creativity, and no dedication.

In the future there'll no doubt be AI DMs, but even if they get very good, unless you're an autist, they'll be a poor substitute for the social experience of playing TTRPGs with friends (short of strong AI and buying sentient robot D&D TTRPG slaves you can keep in your basement). Future of AI in RPGs is more important for video games, not TTRPGs.

Honestly, having the AI come up with ideas for you is asinine to me. You'll never grow as a GM if you do that and every game that relies on it will eventually devolve into a homogenous soup of the same overhashed ideas.

One might make an argument that basically everything is "soup of the same overhashed ideas". It's just how the human brain can mix them compared to an AI. Better so far.

Still i agree that AI is more important and has (hopefully) future in CRPGs not p&p RPG. The only reason i could see for an AI in ttrpg is when no one wants to GM a given system but the group wants to play it. Still, no AI can be made good as human for that purpose, as people said.

 

[0wca]

Only use right now is for generating character and map art. It's hard to think of much use other than art for AI in TTRPGs. Even if you want something advanced and computerised that can do the math for you, all you need is simple scripting for that, not AI, since FoundryVTT can automate all of that already with JavaScript. You can use AI to help with writing and ideas if you're DMing, but besides that taking the fun out of it, you will have better results and more importantly improve your own skills by doing it yourself. Doing it yourself is even more important since probably more and more people will rely on AI to do those things for them with time, meaning there will be a lot of people with zero skills, creativity, and no dedication.

In the future there'll no doubt be AI DMs, but even if they get very good, unless you're an autist, they'll be a poor substitute for the social experience of playing TTRPGs with friends (short of strong AI and buying sentient robot D&D TTRPG slaves you can keep in your basement). Future of AI in RPGs is more important for video games, not TTRPGs.

Honestly, having the AI come up with ideas for you is asinine to me. You'll never grow as a GM if you do that and every game that relies on it will eventually devolve into a homogenous soup of the same overhashed ideas.

One might make an argument that basically everything is "soup of the same overhashed ideas". It's just how the human brain can mix them compared to an AI. Better so far.

Still i agree that AI is more important and has (hopefully) future in CRPGs not p&p RPG. The only reason i could see for an AI in ttrpg is when no one wants to GM a given system but the group wants to play it. Still, no AI can be made good as human for that purpose, as people said.

AI effectively makes something akin to procedural generation. It's making random permutations to make something seem different and unique, and therefore, worth exploring.

However, if you've played any sort of game where areas are procedurally generated, you'll see that sooner or later it becomes all too familiar. Its randomness effectively becomes predictable and therefore, not unique, since it doesn't know why its putting certain permutations together, it just does so.

Now apply this same philosophy to quests and you'll see why this is just a gimmick that will never have any real longevity...at least not for those of us who still want a degree of quality in their games. I'm sure the hordes of conzoomers will gulp it right up.

 

[Bruuce]

Someone made a Foundry VTT module that adds chatbot NPCs to your game.

Link to the module since I don't have enough Princess Points to edit my posts yet.

 

[Gostak]

https://rpgcodex.net/forums/threads/experiment-with-chat-gpt.146924/post-8452535

I think with the current crop of models that consumer can run locally it's getting better.
Models supporting larger context and being less prone to break down (into repetition frenzies or mumbo jumbo etc.) with more.
With stuff like LoRA even fine tuning the by now really huge ANN models still can be feasible offline.

So, using the tech to be an okay-ish DM/GM "substitute" for Dread in a pinch can be had today, I bet!
Love Dread, I actually should try this.
An excuse to look into making my first LoRA.

For the mumbo-jumbo part check this (in this case it still was pretty cool because it was still somewhat quite sensible and more akin to artistic/flowery through-the-roof):
https://cdn.discordapp.com/attachme...afd2eda1779eef1cb63d4428994f97a34f345cee2e1e&

 

[orcinator]

You can use it to make copying text from a PDF much easier, especially when a simple "replace ^p with space" command isn't enough, though you have insist that it doesn't change the words.

 

[anvi]

I think AI will improve everything by 1000x. But there are hardly any good people working on good things, so mostly it will help shit people create more shit to ruin our lives. Also soon the internet will be 100% bots.