🎧 How to Make a Video Game With ChatGPT in 60 Minutes

OpenAI’s Logan Kilpatrick shows us a future where we’re all builders

Every illustration/X.

TL;DR: Today we’re releasing a new episode of our podcast How Do You Use ChatGPT? I go in depth with Logan Kilpatrick, OpenAI’s first developer relations and advocacy hire. As we talk, we build our own text-based strategy game together in 60 minutes using ChatGPT and GPT Builder. Watch on X or YouTube, or listen on Spotify or Apple Podcasts.


You can make a video game without writing a single line of code. Logan Kilpatrick and I did it together live on my show. 

Logan is OpenAI’s first developer relations and advocacy hire. A big part of Logan’s job is supporting the community of builders using ChatGPT, DALL-E, and the OpenAI API. He’s also deeply invested in growing this community, convinced that AI tools empower people to build more. 

To prove this point, Logan and I fulfill a childhood dream of ours, to build a video game, live on the show. We use GPT Builder and ChatGPT to create Allocator, a text-based strategy game where players step into the shoes of a historical U.S. president and are tasked with managing the government’s budget. We have an awesome time iterating all the way from rough idea to functional video game in less than one hour—without any coding. It’s an incredible experiment in how AI can bring our creative ideas to life. 

Three months ago, OpenAI released the application that made our video game experiment possible: GPT Builder. This tool enables people to make custom GPTs tailored for pretty much anything they want. Logan believes that GPT Builder is the biggest technological unlock since ChatGPT was released. It reduces the barriers to innovation and building, especially for people who don’t know how to code.  

This episode is a must-watch for anyone who wants to turn their unstructured musings into  tangible output. Here’s a taste:

  • Expanding the horizon of who can build things. Logan is excited about GPT Builder because it empowers people who don’t know how to code to build custom GPTs for themselves. “[Y]ou can actually make something that is materially more useful than base ChatGPT itself without writing a single line of code,” he says. 
  • Learn how to code! Even though OpenAI is creating tools for people without coding know-how, Logan believes that learning to code is “the highest leverage thing you can do in your life.” AI is already enabling developers to tackle increasingly complex problems both faster and more independently, and he only anticipates this to “exponentially” increase with time.
  • Empowering the next billion coders. As the benefits of coding grow, Logan thinks AI will also teach the next billion people how to code. “[T]here w[ere] not enough computer science educators and resources on the planet to teach the next billion people to code… because we have LLMs, people are actually going to be empowered to go and do this in a way that's personalized and empowering to them specifically,” he says.
  • GPT Builder is trained on the basics of prompt engineering. We decide to build a video game that allows the player to choose a U.S. president and then allocate that government’s budget. On Logan’s advice, we don’t instruct GPT Builder to refine our rough prompt because, he explains, “[W]e can iterate on it once we've actually gotten what [GPT Builder] outputted, but [GPT Builder] does do a little bit of that…prompt engineering magic behind the scenes for us.” The name we choose for our video game is Allocator, inspired in part by my essay on the age of the allocation economy
  • SEO-friendly names for custom GPTs. One more handy GPT Builder tip for your custom GPT to go viral: “[N]ame [the custom GPT] something such that people will actually be able to find it,” Logan advises. To “own the SEO space,” you need to think of a unique name because the “more generic that you make the name…the more polluted it’ll be as somebody is trying to find it.” 
  • Instruct GPT Builder to simulate an expert. As Logan and I get deeper into building our video game, we uncover another prompt engineering nugget. We tell GPT Builder to simulate an expert and come up with a framework for gameplay mechanics because we aren’t technical specialists on the “core mechanics” of games. 
  • Tweaking the instructions underlying GPT Builder. GPT Builder plows right ahead and outputs gameplay mechanics that fill in more gaps than we wanted. To limit this, Logan recommends clicking on the Configure tab to “tweak to those instructions to be like, ‘Hey, maybe give the user some input and then let them go back and forth on some of these things.’”
  • Make with GPT Builder, refine with ChatGPT. We still aren’t thrilled with the gameplay mechanics we get from GPT Builder, so Logan suggests using ChatGPT to refine the custom instructions powering the builder. “Part of the system prompt is to keep these custom instructions a little bit short and a little bit not too tactical,” he explains. 
  • Use ChatGPT to format instructions. We ask ChatGPT to format these instructions so that it’s easy for us to copy and paste them into GPT Builder. Logan’s helpful tip is to specifically instruct ChatGPT to output the instructions as “markdown in a code block,” which deletes the bullet points.

As Logan and I play the game, we jot down a list of things we want to improve. Then we get ChatGPT to refine the instructions and input it back into GPT Builder, until we finally got to a version we were happy with. You can play our game, Allocator—tell us what you think and we’ll do a second episode to modify the game live based on your feedback.

You can check out the episode on X, Spotify, Apple Podcasts, or YouTube. Links and timestamps are below:

Timestamps:
  1. Introduction 00:44
  2. Why learning to code is the highest leverage thing you can do 09:18
  3. AI is empowering the next billion coders 13:40
  4. The first prompt in GPT Builder for our video game 35:58
  5. How to increase the chances for your custom GPT to go viral 39:27 
  6. Prompt engineering tips while using GPT Builder 43:00
  7. How to use ChatGPT in conjunction with GPT Builder 56:13
  8. Ready to play our text-based strategy game! 1:06:33
  9. How to finetune your custom GPT 1:19:44
  10. Why you should build custom GPTs 1:43:12

What do you use ChatGPT for? Have you found any interesting or surprising use cases? We want to hear from you—and we might even interview you. Reply here to talk to me!

Miss an episode? Catch up on my recent conversations with clinical psychologist Dr. Gena Gorlin, economist Tyler Cowen, writer and entrepreneur David Perell, software researcher Geoffrey Lit, Waymark founder Nathan Labenz, Notion engineer Linus Lee, writer Nat Eliason, and Gumroad CEO Sahil Lavingia, and learn how they use ChatGPT.

If you’re enjoying my work, here are a few things I recommend:

My take on this show and the episode transcript is for paying subscribers.


Thanks to Rhea Purohit for editorial support.



Like this?
Become a subscriber.

Subscribe →

Or, learn more.

Read this next:

Chain of Thought

The Knowledge Economy Is Over. Welcome to the Allocation Economy

In the age of AI, every maker becomes a manager

8 Jan 19, 2024 by Dan Shipper

Chain of Thought

Admitting What Is Obvious

I’m a writer—what are you?

26 Sep 8, 2023 by Dan Shipper

Chain of Thought

How Sora Works (and What It Means)

OpenAI's new text-to-video model heralds a new form of filmmaking

1 Feb 16, 2024 by Dan Shipper

Thanks for rating this post—join the conversation by commenting below.

Comments

You need to login before you can comment.
Don't have an account? Sign up!

Every smart person you know is reading this newsletter

Get one actionable essay a day on AI, tech, and personal development

Subscribe

Already a subscriber? Login