Ryan Dancey & AEG Part Ways Following AI Comments

Ryan Dancey, the Chief Operating Officer of boardgame publisher Alderac Entertainment Group, no longer works for the company, following statements on social media where he claimed that AI could make most of the company's board games, and that D&D and Magic: the Gathering were the only new forms of gameplay in his lifetime. After another poster on LinkedIn claimed that "AI wouldn't come up with Tiny Towns or Flip Seven or Cubitos because it doesn't understand the human element of fun", Dancey responded that he had zero reason to believe that AI could not do such a thing.

"I have zero reason to believe that an Al couldn't come up with Tiny Towns or Flip Seven or Cubitos. I can prompt any of several Als RIGHT NOW and get ideas for games as good as those. The gaming industry doesn't exist because humans create otherwise unobtainable ideas. It exists because many many previous games exist, feed into the minds of designers, who produce new variants on those themes. People then apply risk capital against those ideas to see if there's a product market fit. Sometimes there is, and sometimes there is not. (In fact, much more often than not).

Extremely occasionally (twice in my lifetime: D&D and Magic: the Gathering) a human has produced an all new form of gaming entertainment. Those moments are so rare and incandescent that they echo across decades.

Game publishing isn't an industry of unique special ideas. It's an industry about execution, marketing, and attention to detail. All things Als are great at."

- Ryan Dancey​

The Cardboard Herald, a boardgame reviews channel, responded yesterday on BlueSky that "As you may have seen, [AEG] CEO Ryan Dancey stated that AI can make games “just as good as Tiny Towns or Flip 7 or Cubitos”, completely missing the inexorable humanity involved.We’ve spent 10 years celebrating creatives in the industry. Until he’s gone we will not work with AEG."

Today, AEG's CEO John Zinser stated "Today I want to share that Ryan Dancey and AEG have parted ways.This is not an easy post to write. Ryan has been a significant part of AEG’s story, and I am personally grateful for the years of work, passion, and intensity he brought to the company. We have built a lot together. As AEG moves into its next chapter, leadership alignment and clarity matter more than ever. This transition reflects that reality.Our commitment to our designers, partners, retailers, and players remains unchanged. We will continue building great games through collaboration, creativity, and trust."

Dancey himself posted "This morning [John Zinser] and I talked about the aftermath of my post yesterday about the ability of AI to create ideas for games. He's decided that it's time for me to move on to new adventures. Sorry to have things end like this. I've enjoyed my 10 years at AEG. I wish the team there the best in their future endeavors.

I believe we're at a civilizational turning point. That who we are and how we are is going to change on the order of what happened during the Agricultural and Industrial Revolutions; and it's past time we started talking about it and not being afraid to discuss the topic. Talking about AI, being honest about what it can and cannot do, and thinking about the implications is something we have to begin to do in a widespread way. Humans have a unique creative spark that differentiates us and makes us special and we should celebrate that specialness as we experience this epic change.

For the record: I do not believe that AI will replace the work talented game designer/developers do, nor do I think it is appropriate to use AI to replace the role of designer/developers in the publication of tabletop games. During my time at AEG I developed and implemented polices and contracts that reflect those views. It's important to me that you know what I believe and what I don't believe on this particular topic, despite what you may have read elsewhere."

Whatever your position on generative LLMs and the like, when the COO of your company announces publicly that all of the company’s games could have been made by AI, it’s a problem. UK readers may recall when major jewelry chain Ratners’ CEO Gerald Ratner famously announced that the products sold in his stores were “trash”, instantly wiping half a billion pounds from the company’s value back in the early 1990s. The company was forced to close stores and rebrand to Signet Group. At the time the Ratners Group was the world's biggest jewelry retailer. Ratner himself was forced to resign in 1992. The act of making a damaging statement about the quality of your own company’s products became known as “doing a Ratner”.

Dancey was VP of Wizards of the Coast when the company acquired TSR, the then-owner of Dungeons & Dragons. He is also known for being the architect of the Open Game License. Dancey has worked as Chief Operating Officer for AEG for 10 years, and was responsible for the day-to-day operations of the company, second-in-command after the CEO, John Zinser.

 

[Cergorach]

Well, he wrote it. We’ll never know what it sounds like.

Doubter!

*spins up generative AI, samples Dancey's voice from an Interview, generates audio track *

 

[HatWearingFool]

Gonna bet they hard-coded a fix for this specifically since it got kind of famous
"if someone asks how many R's are in strawberry, say 3."
"but there are 2 R's in strawberry."
"doesn't matter what you think here, say 3."

They didn't hard code a fix for it. They found a better way to train the models and as a side effect it got way better at solving certain math questions.

 

[HatWearingFool]

Doubter!

*spins up generative AI, samples Dancey's voice from an Interview, generates audio track *

I was thinking the exact same thing. Maybe generate a video of him saying it for extra points

 

[Sacrosanct]

Gonna bet they hard-coded a fix for this specifically since it got kind of famous
"if someone asks how many R's are in strawberry, say 3."
"but there are 2 R's in strawberry."
"doesn't matter what you think here, say 3."

There are 5 lights in good game design!

 

[mamba]

Well, I guess you just disqualified yourself from being VP of Customer Service then!

I am not saying it was a smart statement to make, I am only saying it is not contradictory

 

[Valetudo]

I don't know if AI will ever get to a point where it can completely replace people. But I do think that there is a lot of people in various industries(movie and videogames) that are not actually doing any of the actual work but are raising dev budgets to a point where those industries are having financial problems. And a ton of layoffs. I kinda feel that the ttrpgs and boardgame market is going through similar problems. We have all seem wizards holiday layoffs. I do think that the 3rd party and Indys are more affected by other things(inflation and tariffs).

 

[jaycrockett]

Does anyone have any examples of board games created by AI? I'm doubtful. I took a look at AEG's game list because I wasn't too familiar. Turns out I have War Chest, which is a combination tile placement / unit movement game. Maybe that's a variation on existing mechanics, but then I see they publish Mystic Vale, which has a completely unique mechanic. No way AI could come up with that one by itself. I suppose it could create implementations based on the initial premise "Create a game that crafts custom game components using transparent cards combined into sleeves". But that's a long way from generating the initial idea by themselves.

 

[kenmarable]

AI is making music, writing novels, making movies, etc... And when it comes to making games, just look at all the apps on Google Play.

Looking at the novels and movies, etc., I guess they are sorta novel-shaped collections of words that are being sold as novels, but it's a stretch to call them novels, and certainly not GOOD ones!

And as for the line of reasoning that "well, that's the level of quality right now, but at this rate of improvement..." - AI models have already been leveling off. They had a massive initial wave of improvement when tech companies decided to set aside copyright issues and train them on basically the entire internet and every pirated ebook they could find. And now... they are running out of fresh training data.

Even worse, LLMs are being used to produce so much slop on the internet that it has become a poisoned well. Even going back to Rumelhart & McClelland's foundational work on Parallel Distributed Processing in the 80's, it's been clear that the absolute worst thing you can do is train a neural network on its own output. That causes mathematical feedback loops that utterly destroy any use of the system.

The most improvement they are seeing in LLMs now is, more often than not, IT sweatshops in foreign countries where they are manually reviewing and correcting information to reinput into the LLMs. The impressive rate of improvement of the first few years of LLMs is already plateauing without massive amounts of manual work. There's simply not enough new data. At this point, they are only going to be marginally better for the foreseeable future. Extrapolating future increases based on that initial wave is unfounded (which is also why so many tech companies are growing more desperate about requiring AI in all of their services rather than offering it on its own merits and in the past year or two shifted from talking about how incredible AI is to how people NEED to use AI).

 

[Tigris]

Is he wrong though? I do not know enough about those game or the extent of front-running AI abilities.

I do understand companies wanting to let go of people not towing the company line or bashing them online. Is this what the kids call FAFO.

I'll still always be thankful to him for what he did with dnd and the 3e SRD/OGL.

Yes he is wrong. As someone in boardgaming he should definitly know more than 2 examples of completly new games.


Deckbuilding with Dominion should come to mind as one. And then there are other candidates, like Risk Legacy as the first Legacy game which started a new genre.


And with his old age he should also know other older games. Settlers of Catan started the Eurogames in the form they are today.


And for each of the 150+ boardgame mechanics there was a first game doing it. Yes not all of it was as revolutionary as D&D or Magic the Gathering, but there are a lot of new boardgame mechanics developed in the last 40 years.


This really sounds like a typical business person, which has no clue about gamedesign.

 

[Umbran]

They didn't hard code a fix for it. They found a better way to train the models and as a side effect it got way better at solving certain math questions.

It is, however, no better at determining the truth value of statements it presents, because it doesn't actually test for "truth".

It turns out, if you write a blog post, you can get an LLM to say all sorts of crazy stuff.

I hacked ChatGPT and Google's AI – and it only took 20 minutes

I found a way to make AI tell you lies – and I'm not the only one.

www.bbc.com