Home
Creators
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

How Long Before AI GMs? (Extended) (Patreon)

Published:
2023-08-19 12:00:05
Imported:
2026-06
Tags:
haven't archived this post yet. have a subscription? use the importer!

Videos

Content

Baron DeRopp of Dungeon Masterpiece joins PDM for a discussion about the impact of AI technology on video games. The audio isn't the best because it was filmed on location at GenCon, but the discussion is fascinating. 

Files

Previews only

AI DMs

Comments

Jeff Krus

That was a great insight into how AI will affect D & D and simulator games..

Bagel-Man

Great video as always PDM, keep up the great work!

Cee-Jay

Great video, thanks. My view is that many people over-state what AI can do at the present moment. That said, I also think we tend to overstate what a DM does and what Creators do. The reality is that our hobby is actually very derivative, borrowing from many of the same sources and repeating the same tropes. Humans are amazing and creative... and at the same time we are all the same and very predictable! There is genuine innovation around the edges of this hobby, but most of what we do is repetitive and very predictable. That's not really a criticism of our hobby, it's more a general statement about humanity and the arts. And how creative is the average player? I don't really know. "I want to poison the apple, then swing on the chandelier and chuck the poisoned apple down the monster's throat" might be hard for an AI to handle well... but is that actually normal, or do must players just attack the creature with little strategy? If the later, AI might do a decent job. And as Baron alludes to in the extra time, AI DM might actually do really cool unexpected things. In Chess for example, computers have change the way the game is played, making moves humans simply didn't used to make. D&D is more complex of course.

David Ford

AI may replace worse than average DM’s? Crap, looks like I’m out of a job.

Cee-Jay

Warning: More Blovation Below. Let me put it another way. Some things tends towards a normal distribution, for example physical traits like height. Other things are NOT normally distributed... income for example. Studies done on employee performance suggest that typically, a huge percentage of the company's value is created by a small handful of employees. Most of us just don't make that big of a difference to an organization's performance, for better or worse. I would bet that DM effectiveness does not look like a normal distribution, it's probably more like employee performance (or like income, for example). What this means, is that an AI that is slightly better than average would never the less be better than the majority of DMs. It wouldn't match whoever the best DMs are (and it might not be remotely close to how good they are), but it might still be better or as good as say the vast majority of DMs. And so what looks like an "average" DM performance to you, might still be an improvement for most people. So for people who play with randoms players on the internet, this could really matter. For guys like me playing in-person with his close friends... maybe it doesn't matter much because a big part of the game is just hanging out. Well, unless the AI is also a better friend and husband than I am, at which point I'm screwed. When that happens, after my wife leaves me, I guess I'll just hang in the basement and play Leisure Suit Larry by myself drinking Tab soda mixed with rum, or maybe TEEM and vodka with a twist of lime.

MrGunn

PDM, very interesting video. I agree, in its current state AI is simply Google on steroids, it uses a natural language model for prompts and presents responses in more human friendly format from data it has scraped from the internet or what ever set of databases it has access to. The age old problem of, garbage data in, garbage date out remains unsolved. AI is NOT intelligent, it cannot discover unknown information, be creative, intuitively synthesize information or question the veracity of information or its premises in the same why humans can. Case in point, I wanted to know the average amount of time it took a medieval peasant to build a daub and wattle cottage. I started researching online, after two hours of reading various sources I had my answer. On a lark, I decided to ask ChatGPT the same question. It took ChatGPT moments to find the answer, it was the same answer I had come up with (no-one really knows :/ ). The difference between ChatGPT's answer and mine is, I had to research and synthesize the answer from the available sources paying close attention to the various bibliographies. In the process I learned something, several somethings, about medieval daub and wattle construction that I could not have learned from ChatGPT. To continue, I asked a different answer about a more obscure topic to which I knew the correct answers, ChatGPT returned dismally incorrect responses, and in some cases just made the shizzle up. In conclusion, AI will not be replacing humans or human GMs any time soon. AI is a great tool for prompts and getting answers to questions that have many correct sources. However, it is dangerous to blindly rely on the machine to give the correct answer, a critical thinking human is still a requirement. Thanks and wishing you a speedy and pain free recovery. MrGunn

MrGunn
>>115579363

@Chris Janak I liked your thoughtful response, one counter-point to your statement: "The reality is that our hobby is actually very derivative, borrowing from many of the same sources and repeating the same tropes." I would repeat the oft quoted, there are only 7 story types and only 12 basic plots, thus all of human literature can be argued to be derivative. Putting these basic plots and stories into the kaleidoscope of human creativity and generating compelling and interesting narratives is what currently sets us apart from the machines, AI cannot do this yet.

Cee-Jay
>>115579363

Thanks for your comment Mr. Gunn. Yes, that 7/12 idea you mention is part of my thinking. I think you get right to the heart of the matter... when, if ever, will AI generate compelling and interesting narratives within these basic story types. When I look at the garbage in most movie theaters these days, I think that AI may not be far from almost equaling that right now! I mean, the writers on strike might not have quite the bargaining power they thing they have. But thinking of the very best of movie cinema, then yes we're aren't even close to that yet. That's what I'm trying to get at.... just how creative are the average DM and table, because that's an important part of when AI DMs become relevant.

Jay
>>115581588

@Chris Janak, I enjoyed reading your insightful comments regarding this topic. This statement below clearly resonated with me as I used to use a lean six sigma program to compile data creating flow charts, pareto distribution charts, in order to streamline processes as part of a global human capital distribution program. You are 100% correct in your assertation. Yeah, just like the movie office space in a way. “What would you say you actually do here?” “Studies done on employee performance suggest that typically, a huge percentage of the company's value is created by a small handful of employees. Most of us just don't make that big of a difference to an organization's performance, for better or worse.” The Pareto Principle, also known as the 80/20 Rule or the law of the vital few, states that for many outcomes, roughly 80% of consequences come from 20% of causes. This principle can be applied to a variety of situations, including business, economics, and quality control. The principle suggests that a small percentage of causes have an outsized effect, and a goal of the 80-20 rule is to identify inputs that are potentially the most productive and make them the priority. I also enjoyed your humorous comment regarding AI replacing you as a friend or husband but couldn’t help being reminded of a recent video by Paul Joesph Watson – THOTPOCALYPSE where some guys seem to be spending time with AI girls online. Some of these guys actually believed they were dealing with a real woman. Link below. Anyway, you make some great points regardless of the bloviation. PJW – THOTPOCALYPSE https://www.youtube.com/watch?v=isafYIg0o3c

Jay
>>115583013

Yep. They are only as good as the data that was input. Not sure how soon they will be capable of critical thinking. To be fair, I know a lot of humans that tend to avoid it.

MrGunn
>>115581588

@Jay Really thoughtful points! I am a big fan of Pareto. Sadly it is the 80% of bad management preventing the productive / creative 20% of us from delivering meaningful value that drives me insane!

MrGunn
>>115581588

@Chris Janak Excellent bloviations! Thanks for your insightful commentary.

Rufus Lynks

Until "computers" are designing their own languages and equipment to support their own network topologies and technologies, all will be based on human analogies. We just lack the ability to be creative in a "different" way. RPGs to me are a social experience, regardless of whether it is a one-person party or a ten-person gathering. Everything else is a tool to support that experience.

Jared Davis

Loved the video. I’ve been interacting with ChatGPT for a few months, sometimes regarding TTRPG design. The main benefit seems to be its ability to scan through massive volumes of content within a few seconds and generate an output. However, I still have to provide it with prompts, guidance, and refinements. It lacks the intuitive spark of experience-based creativity that humans provide. In summary, the current generation of AI tools do what computers are designed to do. They compute at high speed. However, their generative side lacks the inborn imagination that people bring to the table. So, until AI tools have their own 2001 “Dawn of Man” experience with the Monolith, I expect we’ll still be running games for a long time to come.

MrGunn
>>115596637

Battlestar Galactica (TOS 1978) S1E2 Apollo talking to Boxey: Boxey: [The Cylons} They're not smarter! Apollo: In some ways they are, they're programmed to think a lot faster than we do. On the other hand, they are not very individual, we can do a little more of the unexpected, it is about the only advantage we have." By your command.

James C
>>115583013

A bit of a tangent but those who control the input, controls the AI. Think of the implications. If the ones controlling what data can be used (ie which website is acceptable) decides to remove certain inputs, then the ability to manipulate the masses is enormous. We joke about it but critical thinking is being co-opted by the internet and wiki pages. People don’t want to research or discover. Interestingly, physical meeting with friends and playing tabletop in some ways encourages critical thinking. You are given a set of resources and a problem. The solution is your mind.

Jay
>>115583013

@ James C. A big YES to all of that yet they will still blame the AI or the so-called algorithm rather than those who actually provide the input controlling the AI. That is my biggest worry. The tabletop does create critical thinking for people who want to do more than just roll dice and hit stuff. PDM does a good job of demonstrating moral dilemmas as well as political and social issues in his games and how players might improvise to solve problems. Hopefully, others do as well.

samurguybri

What a great discussion. i really appreciate how Baron is finding what AI is good it and leveraging that for the hobby.

samurguybri
>>115581588

Fantastic thoughts, folks! While it may seem that the type of DM doesn’t matter much, I think the huge learning curve and complex set of skills that is required of an even mediocre DM is very impactful and hard to understate. Just think of the skills mentioned in the video and things that we are aware of: following rules and procedures , adapting them to the current state of the game, thinking a few steps ahead, putting oneself in an NPC’s shoes,(tactically, strategically and in their personalities), developing themes, the all elusive pacing, a sense of wonder and tension to name a few. All very instrumental to game play, even if they are not managed well by a new DM, they have to have implement many actions, just to run a very basic game. I think that the Pareto principle may not apply to DnD as much since the game looking to generate a whole slew of hard to quantify effects that all add up to a good time. The good time will vary by group and DM and be session. In that way it’s more like producing and artistic experience than a product like a business does. Sorta! Ha!

Kristen Solindas
>>115581588

You're getting at something I find most people miss: the difference between mean and median. Measurements of central tendency have always been a struggle for humans to properly quantify, leading to skewed perceptions. You lay out a good example of this. Also, I love the Pareto principle, also seen in Zipf's Law, weirdly, as it exemplifies how the human brain is inexorably tied to naturally occurring statistical distributions. It sometimes makes me wonder if we are forever tethered to this way of thinking, and the only exceptions are those that prove the rule. I've played RPGs since 1984, and there have only been a few moments in all this time where I have been truly surprised at the exemplary way the GM weaved a story.

7DSystem

I once had a customer who came into the computer store wanting to buy a solitaire machine. He had a budget of $2000. The salesman in me could definitely have utilised that entire budget, but the human in me said, "Dude, just buy a $1 pack of cards"