An AI trained on Dungeons & Dragons

I have begun development and training on a GPT model , built from scratch. With the ultimate goal of it becoming and aid to players and Dungeon Masters. When ready it will be released under the Open Gaming License. as DnD-GPT. It will be fully open source. 

What do you all think? 

Quote from wagnarokkr >>

Since you asked, to be brutally honest, I think it's an awful idea. Generative AI tools like this are actively responsible for an immense amount of environmental and sociological harm, are usually based on blatant theft of other people's hard work, and have yet to show that they can actually solve any problem better than cheaper and less harmful tools that already exist. Because I love D&D, I think that things like this should be kept as far away from it as possible.

 This. I don't want AI anywhere near my tables.

While I don't agree with your sentiment I thank you for your candor.  The initial data the model is being trained on is in the public domain, and as for the D&D data that  it will eventually be trained on. I have already contacted WOTC and they have already confirmed that this project fits within the terms of their Open Gaming License. I am actually purchasing the Core Rules and not using already pirated materials to acquire the D&D data.

We will just have to agree to disagree then. Every person is responsible for their own actions and any harm that comes to them is generally their fault. An alcoholic knows that they are an alcoholic, it's not the alcohol's fault that the alcoholic doesn't care what it is doing to them.  The same can be said of lots of other things in life. At one point when D&D was new, it got blamed for societal ills and considered the work of the Devil. It wasn't Gary Gygax's fault certain delinquents did bad things and drew attention to D&D which got the blame for their bad behavior. At some point humanity is going to have to take responsibility for its own failings, and stop blaming life's ills on other things that they, themselves created. AI is just the latest in a long line of scapegoats.

Honestly I find the sociological issues of AI to be overstated, though I haven’t looked into the issues of power draw, waste heat and other emissions, and similar ecological concerns. But regardless of all that, everything I’ve seen about Gen AI says that about the only benefit it can contribute to narrative crafting is the speed it spits out words, but its inability to actually hold a concept in mind and create a rational structure mean it cannot contribute as more than a boondoggle to actually running a game. You can’t ask it to plot a narrative, play a consistent character in dialogue, or adjudicate rules- particularly if the rules issue involves any qualitative elements or really requires more than reading a single section from a rule book. It certainly can’t DM and is maybe only slightly more effective than a program that just pulls a set of entries from premade tables like the ones the put in sourcebooks for generating character design/personality prompts.

I think a good AI model could be a very useful tool for a DM when there is a need to quickly describe stuff that hasn't been planned in advance. I'm sure there are people who can come up with a detailed tavern (staff included) on the spot, but that sort of thing shouldn't be a requirement for running a game.

"Even if the environmental issues were somehow magically solved, generative AI tools still can't create anything new, reliably answer a factual question, or do anything that existing tools can't do better. So what benefit do they provide?"

Environmental issues aside for a moment. So your whole objection is that AI is not good enough at this point in time so what's the point? That's it? Let's break this down a bit:
1. AI tools still can't create anything new. Humanity has this very same issue, most things these days are derived from other people's work. D&D keeps evolving based on Gary Guygax's work. Home Brew stuff all based on the intellectual property that belongs to WOTC, the quality of which depends greatly on the individual talents of the creators. But it really isn't anything new, it's a derivative work to solve a particular problem or scratch an itch. I would counter this argument with AI tools still can't create anything new, yet.
2. Reliably answer a factual question. No sure what you mean by a factual question, I mean a question is asked because you either don't know the answer or maybe you want to confirm what you already suspect. As far as reliability is concern that's on the user to check the response for accuracy. Even humans can be unreliable and when it comes to the reliability of AI or any piece of software really it comes down to how reliable are the people that built it?
3. Or do anything that existing tools can't do better. What's your basis for comparison? What tool for D&D exists that can take a natural language input like: Give me a list of monsters with at least 5HD and no more than 9HD and generate an encounter table; then execute the request? 

It seems you expect perfection when we mere mortals are far from perfect. Even the tools we create are imperfect and often have a steep learning curve, where AI only requires the ability to type a coherent and concise input to receive a desired output. And as with most things involving computers, garbage in, garbage out.  

The benefit is clear - D&D has always had a DM shortage.

I'm not advocating for this; I share many of the same concerns. At the same time, AI development is a runaway train and there's not really anyone who can stop it at this point. It's like burning the Amazon rainforest so we can graze cattle for a few years before the soil gives out - any sane, reasonable person would object to it but here we are. So we need to do what we can to call for regulations and guardrails but also be prepared for the big players to do whatever the hell they want to anyway.

As for OP, I think you should take a hard look at what concrete benefits your model would have over a more powerful generalized one, because those exist and people are already using them. The industry is brutal with a lot of billionaires fighting to corner the market and the one that wins is going to try their best to stomp out any smaller, specialized tools like the one you're aiming to make. Or they won't even have to because everyone just uses HAL (or whatever it will be) for everything, including their D&D game.

On a more personal note, I'm not interested in AI assistance in D&D because there's nothing about the process I want to replace. I love using real art for my games and AI images are off-putting to me. I love coming up with storylines and creating monsters. I even love looking up rules during the game instead of just asking a chatbot. Yes, it's hard to come up with detailed descriptive info on the spot, but I believe that putting my brain through those paces is extremely good for my ongoing cognitive function. It's all a part of the game for me. I could see how a new person coming into the hobby may not have the same concept of what D&D is and what it can provide them, but to me AI in D&D feels like getting a robot to play the game for you when playing the game is the whole point in the first place.

Since GPT already has been found to have pillaged from artists and writers, I am not sure that you can actually make an ethical AI based on the model. Thumbs down.

Quote from Duke_of_Vicenza >>

I have begun development and training on a GPT model , built from scratch. With the ultimate goal of it becoming and aid to players and Dungeon Masters. When ready it will be released under the Open Gaming License. as DnD-GPT. It will be fully open source. 

What do you all think? 

 I think that people who announce projects rarely, if ever, finish them.

Your GPT model will be outdated and practically useless by the time you finish training and customizing it.

And then you will find that you will never be able to market it.

I would quit now while you are ahead.

Your observations about AI are all true, for now. A day may come where one won't be able to tell if the person they're talking to online is a flesh and blood human or and AI. If it does let's hope it doesn't decide that humanity is a threat. It take a lot of effort for a DM to create a decent campaign with encounters and such. My goal here aside from learning how to develop AI is to ease some of that burden. I look at AI as a really sophisticated data base with natural language capability. It can help give you ideas if you are experiencing a creative block. Produce the stats, even help polish up a narrative you created. 

I don't know if this project will be practical or not, or if anyone would use it once it's finished,  but the only way to find out is to do it an see what happens. I don't know how to solve the environmental issues, and to be honest it's hard to tell what is or is not true when most everyone is so emotionally charged. I get that they are passionate about certain issues, but no problem is ever solved until cooler heads prevail. But AI is the future and you can't stop the future, it will be on top of us before we know, for good or ill.

Quote from scatterbraind >>

AI development is a runaway train and there's not really anyone who can stop it at this point

Nobody needs to. It's already collapsing under its own weight

Somewhat like blockchain, AI is a solution desperately searching for a problem to solve to justify its own existence and make its investors money

For what it's worth, Deepseek 3's recent release is kind of blowing the doors off the story we've been told about how energy-guzzling AI has to be. There is potential for massive efficiency improvements to be made, they are just not prioritized by the big players in the US. I do believe there's a possible future where AI results in overall reduction in energy consumption, if we can avoid destroying ourselves and others long enough to get there.

If you can privately build an LLM that:

1. Doesn't cause a disproportionate environmental impact in terms of energy and water consumption in relation to its output.
2. Isn't trained on plagiarized data.
3. Doesn't prioritize confidence over accuracy in its responses.
4. Doesn't hallucinate.
5. Is able to provide responses that couldn't be served up in equal capability by a search engine.
6. Can comprehend the scope of D&D just from the SRD.

Well you'll have a tool that not just would be novel to the D&D space, but probably the whole LLM space. This is assuming you're fully building your own model use your own training data.

But if your endeavor doesn't meet all of those criteria, well I personally wouldn't see any use for it, especially on points 1-3. If you're building off a framework from an existing LLM such as Chat-GPT, you'll likely have failed at one or more of the above points from the get-go.

"Since GPT already has been found to have pillaged from artists and writers, I am not sure that you can actually make an ethical AI based on the model. Thumbs down."

That's where you are wrong. GPT (Generative Pre-trained Transformer) is a concept not a proprietary holding. The data any given model is trained on the vast majority of which is in the public domain and by definition can't be stolen. Yes some bad actors have done some bad things, but that's not a legitimate reason to think that all such project are done with stolen data, when there is so much out there in the public domain. This include text, images, and sound all in the public domain. And WOTC already confirmed the project falls with in the Open Gaming License, I asked. So there's no Earthly reason for this project to steal data.

Quote from scatterbraind >>

For what it's worth, Deepseek 3's recent release is kind of blowing the doors off the story we've been told about how energy-guzzling AI has to be.

That's if you assume everything we've been told about Deepseek is accurate. There's no real way to verify it, considering its source

How would it be useful? What would you have it do? These are the questions you would have to answer for people to find value in it. "Be an aid to DM's and Players" an aid that does what? And you need to consider the quality.
Many of the results of AI i have seen, have been atrocious in quality, or have felt off in ways that broke immersion.

The dislike for AI isn't just some neo-ludite impulse, or moral panic. It just has not proved itself to be of any worth to a lot of people. The fact it is constantly being pushed on us when we just want to be able to do our thing and be creative.

Quote from Davyd >>

If you can privately build an LLM that:

1. Doesn't cause a disproportionate environmental impact in terms of energy and water consumption in relation to its output.
2. Isn't trained on plagiarized data.
3. Doesn't prioritize confidence over accuracy in its responses.
4. Doesn't hallucinate.
5. Is able to provide responses that couldn't be served up in equal capability by a search engine.
6. Can comprehend the scope of D&D just from the SRD.

Well you'll have a tool that not just would be novel to the D&D space, but probably the whole LLM space. This is assuming you're fully building your own model use your own training data.

But if your endeavor doesn't meet all of those criteria, well I personally wouldn't see any use for it, especially on points 1-3. If you're building off a framework from an existing LLM such as Chat-GPT, you'll likely have failed at one or more of the above points from the get-go.

 Some of you standards are impossible to know or even achieve.

1 is impossible to know and probably impossible to achieve.

2 is easily achievable since there is plenty of data in the public domain

3 is achievable or at the very least can be mitigated

4 can be minimized but not completely eliminated

5 AI already does this and is oft time easier and faster to use than a search engine

6 Unrealistic and impossible with current technology. Comprehension can only come from consciousness and AI is not capable of that.

Quote from Duke_of_Vicenza >>

I have begun development and training on a GPT model , built from scratch. With the ultimate goal of it becoming and aid to players and Dungeon Masters. When ready it will be released under the Open Gaming License. as DnD-GPT. It will be fully open source. 

What do you all think? 

 Are you here to gain an understanding of what posters on this forum generally think, as your OP implied, or you are you just here to argue with anyone who has concerns about AI?

Quote from wagnarokkr >>

Quote from Duke_of_Vicenza >>

"Since GPT already has been found to have pillaged from artists and writers, I am not sure that you can actually make an ethical AI based on the model. Thumbs down."

That's where you are wrong. GPT (Generative Pre-trained Transformer) is a concept not a proprietary holding. The data any given model is trained on the vast majority of which is in the public domain and by definition can't be stolen. Yes some bad actors have done some bad things, but that's not a legitimate reason to think that all such project are done with stolen data, when there is so much out there in the public domain. This include text, images, and sound all in the public domain. And WOTC already confirmed the project falls with in the Open Gaming License, I asked. So there's no Earthly reason for this project to steal data.

(emphasis mine)

If the vast majority of the training data isn't stolen, the model is still based on stolen data.

 Nonsense. A model built from scratch including the training data taken from the public domain and data used under a legitimate License can not be stolen.