Ai, AppSheet, OpenAi, D&D, tabletop role-playing games
Over the end of year holidays, I had the opportunity to work on an exciting project that combined my love of tabletop role-playing games with the power of AI.
My friend Josh and I used to play D&D together all the time, and we were discussing the many ways in which AI is changing the world when he posed an intriguing question:
"Wouldn't it be cool if you could play D&D with AI?"
I immediately saw the potential for combining the flexibility and customizability of AppSheet with the advanced language processing capabilities of OpenAI to make this a reality.
As the Game Master (or GM) of the game, it was my responsibility to bring the world to life and facilitate the adventures of the players. Josh provided me with an outline for a few key scenes and encounters, but it was up to me to come up with the system that would manage everything. I quickly realized that one of the biggest challenges I would face was creating an ongoing context system for the AI, so it would be able to understand and build upon what had happened in previous turns.
The most powerful model available at the time was "text-davinci-003" with a maximum limit of 4,000 tokens for the text completion endpoint; I knew that if an encounter ran too long, I would run into this limit and the system would start kicking errors.
To overcome this limitation, I had to get creative and come up with a solution that would allow me to condense the information and keep the number of tokens within the limit.
I decided to implement a nested series of summary AI integrations, where after a response for a player's turn was generated, a secondary AI would then summarize what happened that turn. When the entire encounter was over, another AI would generate a summary of of the summaries, essentially writing a concise paragraph or two about the encounter. Then, once a scene was over, another summary was created of all the encounters that had taken place.
This process allowed me to keep track of the ongoing context of the game and ensure that the AI had all the information it needed to provide meaningful responses. It was a complex system to set up, but it ultimately paid off and allowed the game to run smoothly.
The project was a huge success, with a live stream on the Friday before Christmas that was well-received by the audience, and a ton of fun.
I even created a public app that allowed viewers to follow along in real-time, exploring the responses generated by the AI and suggesting actions for the players.
The app was tied into the database from the game, so viewers could see what was happening as it happened and get a sense of the full context of the game.
In addition, the app had a feature that allowed viewers to submit their own ideas for actions during a turn, which was wildly popular and added an extra level of fun and interactivity to the experience.
Explore the public companion app
The combination of the live stream and the public app allowed for a truly interactive and immersive experience, with people coming up with some of the most entertaining and creative ideas.
The success of the project has left me eager to explore more possibilities and ideas for using AI in new and innovative ways.
The experience of working on this project was eye-opening and gave me a glimpse into the future of work. With the help of AI models like ChatGPT and Codex, I was able to take an idea from my head and turn it into a functioning prototype in just two hours.
In two hours I was able to take an idea in my head, and turn it into a fully functional prototype!
With the ease of use of platforms like AppSheet & Google Sheets, I was able to quickly build a complex, relational database system that facilitated all the data storage and complex interactions. The speed and efficiency with which I was able to accomplish so much was truly impressive, and it's clear to me that AI has the potential to revolutionize the way we work.
It's exciting to think about all the ways in which AI can be used to boost productivity and make our work lives easier and more efficient.
With the success of this project, I'm excited to continue exploring and developing new ideas for using AI in innovative ways. I have a ton of plans for add-ons for AppSheet that will allow for easy AI integrations, such as email text generation, document summaries, executive summaries, key takeaways, data point extraction, and more. I believe that the possibilities for AI integration are endless, and I'm eager to see what the future holds.
I've made my generic, universal text and image generation scripts available to my Patreon supporters (www.Patreon.com/MultiTech).
For just $10 a month, my supporters have access to a wealth of resources and tools that I've put together over the years as an AppSheet consultant, all designed to make my life easier and help me work more efficiently.
If you're interested in learning more about what I have to offer, I encourage you to check out my Patreon page and see all the resources I provide.
*All of the images in this post were generated by Ai. Images in this post © Story-Time with Ai by MultiTech Visions. All rights reserved.
Sharing with credit permitted
My project exploring the idea of "Story-Time with Ai" was a fantastic learning experience that showed me the incredible potential of AI to boost productivity and facilitate new and innovative ideas. From the success of the live stream and the public app to the endless possibilities for AI integration, I'm excited to see what the future holds and to continue working on new projects that push the boundaries of what's possible.
Thank you for following along on this journey with me, and I hope you've enjoyed reading about my experience as much as I enjoyed creating it.
Oh... and PS: this blog was written by an Ai from an outline I provided.
Start to completion took less than an hour. (+5 min) for the outline (+5 min) for some revisions (+30 min) for formatting the post How long does it take YOU to make a blog post!? (good stuff coming down the pipeline soon...) - Have a nice day! (^_^)