using chatGPT to make your own games
"Bernie and the Samurai" Episode 1.
It is 2025.
You and your spouse care sitting down to watch some serious entertainment.
You ask, “So what do you feel like?”
Your spouse says “how about Bernie and the Samurai?”
This is a favorite. You and your spouse invented it. It features Bernie Sanders palling around Elizabethan London with a Japanese Samurai.
It’s pretty fun. You invented it, sure, but chatGPT did most of the work.
After some programming preliminaries (as below), you said:
Please create a script and a setting for Bernie Sanders and a Japanese samurai as they find themselves negotiating life in 16th century London.
Here’s what chatGPT wrote for you.
Bernie Sanders: Dressed incongruously in a simple but clean tunic and breeches, trying to blend in but still noticeably foreign with his thick Brooklyn accent.
Takashi: A stoic Japanese Samurai, wearing a nondescript cloak over his armor to avoid drawing too much attention, his katana discreetly hidden.
Setting
Location: A bustling market square in 16th century London, filled with merchants, townsfolk, and the distant sound of carriages on cobblestone streets. The atmosphere is lively yet tense, with a mix of various cultures due to increased trade.
Time: Mid-afternoon, with a light drizzle adding to the mud and chaos of the market square.
[Scene opens with Bernie and Takashi walking through a market, observing the various stalls and goods.]
Bernie: (gesturing to the chaos around them) You know, Takashi, back in my time, this would be considered quite the regulatory nightmare. How are you finding London so far?
Takashi: (looking around cautiously) It is... different from Kyoto. More noisy, more dirty. But there is strength here, too. Different, but strong.
Bernie: Exactly! It’s the strength of the people. See, back where I come from, I always argue that the strength of a society lies in its fairness. Here, I think, we could do a lot of good.
[A commotion starts a few feet away from them, where a local thief has snatched a loaf of bread and is running away. A guard starts chasing him.]
Takashi: (nods) In Japan, honor is our way of life. Dishonor, like stealing, brings shame not just to the individual, but to the whole family.
Bernie: And that's a good point! Here, people often steal out of necessity, not dishonor. If only they had enough to begin with... Ah, it's all about economic justice, my friend.
…
[They reach the end of the market, facing the Thames, a view of ships loading and unloading goods.]
Takashi: (looking out over the river) So, Bernie, where to next in this vast city?
Bernie: I say we find ourselves a decent pub and talk about healthcare. I've got some ideas that might just work, even here.
[Both laugh as they walk off towards the river, the camera pulling away to show the sprawling city of London.]
Oh, yes, listening to Bernie talking about healthcare in the plague ridden, regulatory nightmare that is early modern London, that’s going to be a beaut.
Let’s make that episode 2. Bound to be riveting.
So where did this idea come from?
Yesterday I was listening to the A16Z podcast.
The topic was next-gen gaming.
The host was Steph Smith and the guests were Andrew Chen and Josh Lu.
The theme was creating games that “integrate seamlessly with the digital age” and “move at the speed of culture.”
Last night, I asked myself “Hey what if chatGPT could work as a game grammar.”
So this morning I asked chatGPT:
Grant:
If I gave you 3 paradigmatic Classes each with 4 items, could you assemble 1 item from each class to create a series of sytagmatic chains?
chatGPT:
Certainly! Please provide the three paradigmatic classes along with their four items each, and I'll help you assemble them into syntagmatic chains.
Grant:
Class one should include these items: Jerry Seinfeld, Margaret Atwood, Bernie Sanders, Stuart Brand; Class two should include these items: wild comedy, deep tragedy, dark noir, tender romance; Class three should include these items: 16th century London, pre-colonial Brazil, china before the emperors, imperial Japan. please create 10 chains.
chatGPT:
Here are ten syntagmatic chains, each combining one item from each of your provided classes:
Jerry Seinfeld - Wild Comedy - 16th Century London
Imagine a sitcom scenario where Jerry Seinfeld's comedic sensibilities clash with the Shakespearean actors of the time.
Margaret Atwood - Deep Tragedy - Pre-Colonial Brazil
A narrative exploring deep human emotions and tragic events in the lush, unexplored forests of Brazil before colonization.
Bernie Sanders - Dark Noir - China Before the Emperors
A film noir style story where Bernie Sanders is a detective uncovering corruption in ancient Chinese states.
Stuart Brand - Tender Romance - Imperial Japan
A love story set in the courts of Imperial Japan, infused with themes of innovation and cultural preservation.
Jerry Seinfeld - Tender Romance - Pre-Colonial Brazil
A romantic comedy where Jerry navigates the complexities of love in the vibrant and untouched landscapes of Brazil.
Margaret Atwood - Dark Noir - 16th Century London
A dark, atmospheric tale penned by Atwood, set against the backdrop of plague-ridden London.
Bernie Sanders - Wild Comedy - Imperial Japan
A humorous, yet insightful look at cultural exchange as Bernie navigates the intricacies of Japanese politics and society.
Stuart Brand - Deep Tragedy - China Before the Emperors
A tragic story exploring the loss and legacy of ancient wisdom as civilizations rise and fall.
Jerry Seinfeld - Dark Noir - Imperial Japan
Seinfeld as a cynical detective uncovering mysteries in the shadowy corners of samurai-era Japan.
Margaret Atwood - Tender Romance - China Before the Emperors
An epic romance novel set in the mystic and historical landscapes of ancient China, crafted by Atwood's masterful storytelling.
chatGPT concludes: These chains creatively merge each item from your classes into compelling and imaginative scenarios across different genres and settings.
No, it’s not perfect. But it’s a first try, provoked by simple questions and dispatched in 20 seconds.
A next stage beckons. We want to give chatGPT access to our blog posts and social media, and anything else it can find, to craft a character called “Grant.” And then it will insinuate him into the story. Personally, I would love to see what happens to me as I tag along behind Bernie and the Samurai. You know, in the mid-afternoon, with light drizzle adding to the mud and chaos of 16th century London.
And there is a ton of creative tinkering to be done here. (Why should chatGPT have to do all the work?) What happens when we take these too predictable terms, wild comedy, deep tragedy, dark noir, tender romance and stir them a little: wild tragedy, dark comedy, deep romance and tender noir. The script begins to breathe a little.
And that’s the end of entertainment as we know it. Who will want to sit down to watch a story written by someone else about someone else? It will all seem so very “silent picture.”
There's a lot of this in the popularity of the first two Star Wars: Knights of the Old Republic games.
That sense that you are in the story, making decisions, choosing your path, not only through the adventure, but also between Light and Dark. Of course, many other games in other settings work in a similar way with varying levels of success.
I think what will be interesting technology wise is the development curve for: narrative arc, character development (which are both linked into plot) goes - as this is where we get into meaning territory - and there's a lot of evidence that on the one hand (cow clicker) people can make meaning out of anything, but at the same time, the nudge that great writing brings around meaning has a lot of value for a lot of people and will take some development.
Finally, sometimes we'd rather events were happening to other people (so we can draw meaning from it, in a slightly distanced way.)
Grant, this was a fun introduction for me. Delightfully creative.