Universal Adventure Simulator

This project dynamically generates adventures that can take place in any setting or genre, from any point of view, that can go on forever in any direction. Want to play as a seagull in a noir film? Want to destroy Ancient Rome as a Cyborg Dinosaur? What about playing as a forest, or an amoeba? Want to work a farm in a fantasy universe, or navigate the inter-tribal dynamics of a clan of jungle cats? Boot up the Universal Adventure Simulator!

Here’s what the device looks like in-person (I can’t show you the real thing of course, it’s obviously top-secret)

Here are some example adventures:

How’s it work?

This project uses GPT from OpenAI as an interactive storyteller. When you say what you want to do, an AI continues the story in response, similar to playing a roleplaying game with a friend.

Does it actually work?

This system has limitations! It doesn’t have a perfect memory - it can forget details like the weather, some plot points, and some character traits. GPT can’t actually generate random numbers, and in the current system, it just makes one up. Responses can dip into generic or monotonous spaces, especially in boring contexts. And, compared to other games, it’s computationally expensive to run.

What does the Universal Adventure Simulator do well?

Where it excels is in breadth and versatility. I tend to feel most impressed when exploring scenarios that are really unique and strange, like playing a game as “the wind” or taking wild left turns and seeing what happens. In a sentence, the more you’re engaged, the more engaging the story will tend to be.

How can I try it?

You can play now for free with the links above or below!

What’s next for this project?

Some of the limitations may be solved soon by the existing trajectory of AI and hardware development. Other limitations require more active development, like more engaging story frameworks, better memory systems, multiplayer experiences, and so on. If there’s enough interest in the current state of this project, I see a few ways these systems can become more advanced. If you want me to keep working on this, let me know you’re a fan!

buy me a coffee!

You can also support me by buying things from my shop: