A downloadable book

Download NowName your own price

The Real World 

The first  playable TTRPG written by AI!

The Real World is a rules lite tabletop roleplaying game written by AI set in a delirious hellscape of whimsical horror. 

Earth, as it turns out, is actually just The Cave, a small and boring pocket of the hyperdimensional chaos of The Real World.

 Hieronymus Bosch, Clive Barker, and  Terry Gilliam, are a few of the inspirations for this setting, among many others.

Each of the players act out the role of one particular individual. This might be a human from The Cave, or a denizen of The Real World. Literally anything can be an ability in this system. Whenever you attempt something, you declare the name of the ability you are using, and roll a number of six sided dice.  If none of those dice are 5s or 6s, that named ability levels up. 

This 72 page book   contains the rules set,  a description of 20 factions, 100 strange beings, 100 strange places, and 100 strange objects.

You can also get this book print on demand from Amazon:  https://www.amazon.com/dp/B09HR81V61/ref=cm_sw_r_apan_glt_i_80BHHEFCA3M8PMVZD8SY

Note: Contrary to the back cover, this book was written using not just GPT-3 via Sudowrite, but Meta AI Megatron-11B as well.

Release date Oct 10, 2021
Rated 4.3 out of 5 stars
(3 total ratings)
AuthorAlley Wurds
GenreRole Playing
Tagsartificial-intelligence, Experimental, Surreal
Average sessionA few hours


Download NowName your own price

Click download now to get access to the following files:

The Real World 46 MB
The Real World no art for strained eyes.pdf 279 kB


Log in with itch.io to leave a comment.

If you're wondering where to find people who play such strange rpgs, I run an LFG discord server where generic fantasy and generic space opera are banned. 


As some readers found the original art of the book distracting from the text, I have uploaded a document called The Real World no art for strained eyes. It's just the text of the book, so that it will be easier for some to read. Thanks for the input!


Let me know if you have any questions!


Could you talk about your process for training the AI?--if "training" is even the right word!



I didn't train it myself, though it does start to use the book in progress as context once it gets going.

I used a combination of sudowrite and inferkit. Sudowrite was better for bulk generating the entries that make up most of the book, and inferkit was better at fleshing out those entries with more detail.

Most of my labor was curation, as I threw out about 95% of what GPT-3 generated. This is important, as what you keep influences what gets generated next.

Hopefully that answers your question, but feel free to ask more!

Cool, thanks!