Podcast #256: Algorithmic AI Content Generation and RPGs

This week, Morrus and Peter talk about algorithmic AI generated content and RPGs. In the news, D&D Beyond, Itch.io, Roll20, and Fantasy Ground summer sales, official Adventure Time RPG announced, Baldur’s Gate 3 release date change, and more! Plus a listener question about Creative Commons vs ORC License and brand new sketch about interesting alchemical experimentation.

Subscribe on iTunes | Google Play | YouTube | TuneIn | Stitcher | Blubrry | Podbean | Spotify | Android

#256: Algorithmic AI Content Generation and RPGs | Morrus’ Unofficial Tabletop RPG Talk

This week, Morrus and Peter talk about algorithmic AI generated content and RPGs. In the news, D&D Beyond, Itch.io, Roll20, and Fantasy Ground summer sales, official Adventure Time RPG announced, Baldur’s Gate 3 release date change, and more! Plus a listener question about Creative Commons vs...

morrus.podbean.com

​

News

D&D Beyond Summer Sale https://www.enworld.org/threads/theres-a-big-d-d-summer-sale-on-right-now.698639/

Itch.io Summer Sale itch.io Summer Sale

Official Adventure Time RPG Announced from Cryptozoic https://www.enworld.org/threads/official-adventure-time-rpg-uses-new-yes-and-system.698381/

News Digest for July 7 (Baldur’s Gate 3 release date changes, Blood in Baldur’s Gate prequel, Call of Cthulhu events with Mark Meer, The One Ring $2 million Magic: The Gathering card found) https://www.enworld.org/threads/news-digest-for-july-7.698630/

What’s OLD is NEW (WOIN) Starter Box Kickstarter launches https://www.enworld.org/threads/wha...ern-sci-fi-ttrpg-with-an-open-license.698590/

Gloomhaven BackerKit campaign over $3 million Gloomhaven Grand Festival

Savage Worlds 20th Anniversary Celebration Kickstarter https://www.kickstarter.com/projects/545820095/savage-worlds-20th-anniversary-celebration

BREACH! Public Playtest on Itch.io BREACH! Public Playtest by Orange/Indigo

--------------------

Algorithmic Content Generation and RPGs

OpenAI Hit With Class Action Over ‘Unprecedented’ Web Scraping from Bloomberg Law OpenAI Hit With Class Action Over ‘Unprecedented’ Web Scraping

Authors file lawsuit against OpenAI for unlawfully ‘ingesting’ their books from The Guardian Authors file a lawsuit against OpenAI for unlawfully ‘ingesting’ their books

US Copyright Office: AI Generated Works Are Not Eligible for Copyright from ARTnews US Copyright Office: AI Generated Works Are Not Eligible for Copyright

Thought experiment in the National Library of Thailand
The Chinese Room Argument (philosophical experiment about a person alone in a room following a computer program for responding to Chinese characters slipped under the door) The Chinese Room Argument (Stanford Encyclopedia of Philosophy)

Chinese Room 2 (a spin-off of the Chinese Room Argument from Tumblr user lumsel to explain how ChatGPT and other large language models work) chinese room 2

Lawyer Used ChatGPT In Court - And Cited Fake Cases. A Judge Is Considering Sanctions from Forbes Lawyer Used ChatGPT In Court—And Cited Fake Cases. A Judge Is Considering Sanctions

Dungeons & Dragons Player Gets AI Bot to DM, Works Surprisingly Well from TheGamer Dungeons & Dragons Player Gets AI Bot To DM, Works Surprisingly Well

American Scientist article on Bias Optimizers Bias Optimizers

ChatGPT gender discrimination example with nurse and doctor, doubles-down on sexism by stating that the doctor must be male and nurse must be female from April 2023 Hadas Kotek: Doctors can't get pregnant and other gender biases in ChatGPT

ChatGPT’s “jailbreak” tries to make A.I break its own rules or die from CNBC ChatGPT's 'jailbreak' tries to make the A.I. break its own rules, or die

Artists Voice Concerns Over the Signatures In Viral LensaAI Portraits from ARTnews Artists Voice Concerns Over The Signatures In Viral LensaAI Portraits

Getty Images is suing the creators of AI art tool Stable Diffusion for scraping its content from The Verge Getty Images is suing the creators of AI art tool Stable Diffusion for scraping its content

Gizmodo’s io9 Published an AI-Generated Star Wars Article That Was Filled With Errors from Variety (it was a list of Star Wars content in chronological order that left out multiple shows including Andor and Kenobi and lists The Clone Wars coming after Rise of Skywalker, which is incorrect both in-universe and in real-world release dates) Gizmodo’s io9 Published an AI-Generated Star Wars Article That Was Filled With Errors

A sci-fi magazine has cut off submissions after a flood of AI-generated stories from NPR
ChatGPT uses between 1L of water per 40-100 queries ChatGPT needs to 'drink' a water bottle's worth of fresh water for every 20 to 50 questions you ask, researchers say

--------------------

Please support us on Patreon at http://patreon.com/morrus

Don’t forget to join the Morrus’ Unofficial Tabletop RPG Talk Facebook group at https://www.facebook.com/groups/1033145023517295/ and join us on Discord at https://discord.gg/VAuxX8M

Ask your Listener Question on Twitter, email morruspodcast@gmail.com, or contact us on TikTok at TikTok

--------------------

Hosts: Russ “Morrus” Morrissey, Peter Coffey, and Jessica Hancock

Editing and post-production: Darryl Mott

Theme Song: Steve Arnott

Malach the Maleficent played by Darren Morrissey

Check out all the media content from EN World at http://enliverpg.com

 

[Egg Embry]

This is an important conversation to have. Thanks for taking the time to talk about AI.

 

[Emberashh]

My 2 cents on the matter has always been that if the AI cannot function without the use of content its creators were not explicitly authorized to use, then it doesn't matter whether or not any of that data makes it back out onto an output. If those outputs are being commercialized, it's IP theft, plain and simple.

And thats without getting into the fact that AI's, unlike humans, can't improve upon their outputs in a vaccuum. An AI fed nothing but what they spit out will cannibalize itself into nonsense. A human has the capability to judge and improve their own creations independently of any other influence, and can in turn produce better pieces on subsequent attempts. Thats what sets actual art apart from the glorified stock imagery and lorem ipsums of AI.

But, beyond all that, AI as said is just glorified stock images and lorem ipsums. Good fodder to manipulate into actual art, but its only ever going to be good at replacing the bottom of the barrel, as harsh as that might be to say, without the hand of an actual artist.

Of course it can't be ignored either that AI has already had about as much a negative impact on art spaces as it has a positive one.

Regardless of how you feel about AI, i can't fathom how anyone could see it as a good thing to see forums and image hosts getting spammed 24/7 with garbage AI submissions pulled direct from output.

 

[Abstruse]

As you can tell from the show notes, this is a topic I've been following closely and had most of those links handy. About the only thing that didn't get brought up is that a lot of image generation algorithms (I hate calling them "AI" because it makes people think it's Data painting them a picture of Skynet or HAL-9000 plotting to take over the world) are starting to hit a feedback loop where they're scraping so much artwork without checking credits that they're pulling art from other algorithmically generated sources. So the bot-art is getting trained on other bot-art and picking up bad habits from one another like the Habsburg Chin of image generation.

 

[BrassDragon]

Although it probably won't catch on, I like the way you're framing the technology as a content generation tool, rather than a humanlike intelligence.

To consider the impact on our hobby, we really don't need all the philosophical baggage that is taking us into unproductive debates (like, for example, the Thai library thought experiment - facinating to be sure, but wholly irrelevent to the actual problems and opportunities in front of us.)

Fact is, having readily available technology that can turn sparks of an idea, incomplete thoughts and quick brainstom sessions into actual, viable and enjoyable content (like, say, a whole new game setting, a heartbreaker system tailored to your table, 3D-printed miniatures that perfectly encapsulate your characters, endless additional content for the games you're already playing etc.) is AMAZING.

Everybody is getting worked up about the output, and whether the quality is acceptable, 'soulful' or properly sourced / accredited... but the real revolution is happening on the input side - I don't think people realize how powerful it is to just tell a machine to produce something, in your own words and language. It's going to turn dabblers into creatives and really talented people into content-producing gods.

Yes, right now, the output leaves a lot of be desired. But that's going to change so rapidly. Our current worries are going to seem quaint in a year or two. For all the boycotting and deplatforming of 'AI'-content, I'm absolutely convinced that we'll still have extremely talented and unique game designers in the future... and nearly all of them will be using generative technology in one way or the other, from playtesting to layout to probability table design etc.