The game industry – and the world in general – continues to grapple with the implications of increasingly-impressive generative AI technology. I touched on the intersection of AI and game development in earlier articles, but since then the tech has only become more prevalent and mainstream.
Chatbots are now the first line of defense for customer service departments. Disturbing and misleading “deepfake” images are popping up everywhere – from pornography to the political sphere – and are increasingly difficult to distinguish from the real thing.
Human-authored content on the internet is drowned out by an avalanche of sloppily-written AI articles. Investor money continues to flow into AI startups from venture capitalists afraid they’ll miss out on the next wave of easy profit.
The hype around AI tech is simultaneously overblown and frightening. Existing technology is powerful, though merely slapping AI into existing systems is not the immediate game-changer many investors think it is. While well-intentioned start-ups pursue narrow, practical uses for AI, much of the big money still goes to companies making larger – but fundamentally emptier – promises.
At the same time, proponents of AI are dismissive of concerns about the implications of how training data is collected. The leadership behind some of the biggest and most well-known AI products act like they’re entitled to unlimited access to all of human history’s intellectual property – perhaps because without that access, their business models would collapse.
Lawyers and politicians are struggling to keep up. In the past, our elected representatives have utterly failed to understand the ramifications of comparatively simple social media platforms like Facebook and TikTok. Pity the poor fools as they dive into the far-murkier waters of generative AI technology.
So it’s a dangerous moment. While I believe a fairly heavy-handed regulatory check needs to be applied (and soon!) to the Silicon Valley bros pushing their vision of completely unfettered AI technology, game developers who try to remain “pure” and ignore what’s happening in the field do so at their peril.
Because even tentative dabbling with existing tools will quickly show how much value there is – and how, rather than replacing humans, proper application of generative AI speeds up and improves the creative process.
![](https://i0.wp.com/screegames.com/wp-content/uploads/2024/06/OldAlexa.jpg?resize=1024%2C768&ssl=1)
home, including this dusty old thing.
Robots (and Amazon!) are listening…
AI as Forever-DM
Two weeks ago, I wrote about Dungeons & Dragons (again – it’s a topic in regular rotation on this blog). The most famous tabletop RPG system in history remains one of my primary influences – a pastime that ate up months of my childhood and was responsible for making me want to become a game developer.
D&D is a time-consuming hobby. It’s a lot of bookkeeping, prep time, printing out player handouts, and planning encounters. Even for players who never DM, there’s a lot of reading and writing and erasing and scribbling – and creative thinking.
Modern apps smooth out the process of playing tabletop RPGs, but the more I play D&D on a platform like Roll20.net, the more I wonder why I don’t just replay Baldur’s Gate 3 for the fourth time. For me, an electronic element takes some of the fun out of the hobby.
The true addicts among us have an eternal dream of playing solo for times when a group can’t get together. I tried when I was young (oh how I tried!) I’d roll up random dungeons using the extremely limited tables in the back of the Dungeon Master’s Guide, but that mostly resulted in the kind of soulless and flavorless experiences found in the worst of procedural roguelikes.
Today’s solo roleplayers have better options than a few random tables tucked in the back of a dog-eared Dungeon Master’s Guide. Solo-focused systems like Ironsworn and the excellent Mythic Game Master Emulator are great at providing narrative prompt options that help spark the imagination. Indie pen-and-paper games like Ker Nethalas add more context and story to random dungeon crawling.
And then there are AI tools, including some specifically designed to emulate a complete D&D experience for a solo player.
After reading reviews of various apps, I decided to try something simpler – using ChatGPT to supplement running a solo D&D campaign. The results were great. In many ways, serving as an impromptu dungeon master is the perfect use for AI tech.
I told ChatGPT about my character and the party; I gave it a few details about the setting. I fed it some prompts with key things I was interested in, following a format suggested by the excellent Mythic Game Master Emulator system.
In return, I got a great plot hook, a fleshed-out initial village and tavern (complete with several NPC characters), and a first dungeon quest. If I didn’t like a detail, it was easy to re-prompt and make changes. While I’d still take a human DM and other human players any day, AI filled in the gaps “enough” to make for a meaningful and substantive solo RPG session.
![](https://i0.wp.com/screegames.com/wp-content/uploads/2024/06/DDTools.jpg?resize=1024%2C768&ssl=1)
excellent Mythic Game Master
Emulator, and a laptop for ChatGPT.
Sketchier was my use of AI artwork for portraits of characters and images of places. I dabbled with Midjourney, and the results were good. With a few prompts, I got exactly what I wanted for my campaign, replacing my old-school process for finding inspirational artwork – internet scouring of public-domain fantasy portraits to get a “close enough” match.
Legally, the non-commercial use of a Midjourney portrait for a home solo D&D campaign is a pretty clear case of fair use. But I’ll acknowledge that ethically, using AI-generated art at all is on shaky ground.
If I weren’t dabbling with Midjourney, would some human artist have been able to sell me a fantasy supplement of hand-drawn images for my homebrew solo RPG campaign? Would I have been willing to pay for it, and how much? I don’t have a definitive answer to those questions, but they’re valid.
Using Midjourney art for any commercial venture remains fraught with peril while the legal and regulatory space lacks clarity. Midjourney itself recognizes the risks, making a recent change to its Terms of Service which essentially attempts to pass the potential copyright liability buck to its users.
Habitual Line-Stepper
In using Midjourney images as inspiration for a homebrew solo D&D campaign, I’m straddling a fuzzy ethical line, but one that’s probably acceptable to most. It’s AI-generated content used for an entirely non-commercial purpose, stuff that no one will ever see.
My overall creative journey and the fun I’m having aren’t dependent on specific pieces of artwork. My solo campaign is inspired by Midjourney’s generated fantasy portraits in much the same way that the creators of Dungeons and Dragons – and countless fantasy novels since – were inspired by Tolkien’s novels.
But working developers leaning on generative AI tech are crossing the line all the time. More and more cheaply-made shovelware with AI-generated art is vomited out every day. Chatbots go off the rails regularly. New holes are found that suggest the tech’s foundations are wobbly at best.
In a previous blog, I touched on how, as a solo developer, ChatGPT was fantastic for helping me solve simple programming problems. I’m not an engineer, and I thrash around in Unity a fair bit. ChatGPT served as a more powerful Google search, helping me quickly get over the hump on bits of C# code that otherwise would have held me up for days.
AI art from Midjourney has also proven inspirationally valuable. My current budget for my project is literally “my labor and sweat equity.” I would love to pay for an actual concept artist – I have in the past, and I will again in the future. Working with a real artist is a far superior experience; I’ll be able to get exactly what I want in a way that Midjourney can’t deliver.
But as I explore prototypes and iterate ideas, using Midjourney for initial quick inspiration has been a valuable exercise – it feels like a better version of how I used to scour the internet looking for reference images to provide to concept artists.
![](https://i0.wp.com/screegames.com/wp-content/uploads/2024/06/MidjourneyCharacter.jpg?resize=1024%2C893&ssl=1)
Midjourney for my D&D campaign.
Not perfect, but fine for my purpose.
The Debate Isn’t Over
I share the concerns of my artist friends in the game industry. Generative AI is changing the face of game development in unexpected and disruptive ways. Still, there’s no denying the power of generative AI tools to accelerate the creativity and productivity of less technical and less artistically inclined folks.
Some days, I wake up and wish generative AI had never bubbled up out of the Silicon Valley tech-bro swamp. On other days, when I’m stumped for the next steps on my family D&D campaign and need inspiration, a quick ChatGPT session can unblock my creativity in a way that hours of internet searching can’t.
We all need to stay engaged in the conversation and continue to advocate for our concerns and the rights of developers and artists. One framing about generative AI I’ve heard is that we don’t want AI to make our art so we can do our laundry; we want it to do our laundry so we can make art.
I keep coming back to that premise as a great foundational principle.
I think of all the small but critical tasks typically foisted off on junior designers – coming up with a list of names for characters, generating a localization spreadsheet, plotting out a curve of balance numbers, or calculating consistent costs for gear.
They’re the kind of unglamorous tasks that aren’t particularly creative but still have to get done. And they’re the perfect vectors for AI to help with – provided the work is guided and edited by (well-paid and ideally unionized) human hands.
For many game developers, even discussing the topic of AI is anxiety-inducing. It’s a hard time out there in the industry; people need to eat and pay rent. The idea that our careers are imperiled by AI tech makes folks want to run screaming into the hills.
But designers, programmers, and artists who ignore generative AI do so at their peril. Sticking our heads in the sand won’t make the technology go away; the core ideas are too powerful.
We need to prepare ourselves for a world in which – in the fashion that capitalism historically tends to do – current concerns about intellectual property rights and the potential dangers of AI technology might get swept aside by powerful corporations in pursuit of the almighty dollar.
And we all need to be active advocates for what shape the world that replaces this one will take.
New Scree Games blogs appear on Tuesdays, with a repost on Medium every Wednesday. If you enjoy the articles, please share them!