Why we’re not starting with Text Based RPG AI?
Generative AI will significantly change the gaming landscape. This is agreed upon by majorly everyone involved in the gaming industry with the exceptions of Nintendo and a few small studios. The fact that very soon every pixel we see will be generated is also a non uncommon view among the ones who actually are working on the intersection of AI and consumer experiences. So naturally AI generated games or real time generated gaming experience is an exciting idea for many. I align with these people very much. The differences are however remarkable in the path that different people believe that will lead to this mutually agreed upon future.
When we started researching Text To Game AI 8 months ago. Arcane, AI Dungeon and FRVR were the only reasonable names worth talking about. I rejected AI Dungeon and Arcane's route at first glance because I have never been big on Text based RPGs. 2 months back, a creator and a friend convinced me to give this a try. As we're working closely with specific creators, we decided to help them make a Text based RPG. It ended up being a weekend long project. We made it and shared it with a few friends of those creators to get feedback on it.
Our assumptions were:
1. The people who have played Dungeons and Dragons style games want to explore Text Based RPGs powered with AI.
2. The people who have played text based RPGs in the past want to play more of them and creatively experiment with them.
_edited.jpg)
When we shared the proto with them we expected the feedback to be more around how the Text based RPG experience would be better. Expected to hear things like:
1. The randomness needs to be increased/decreased.
2. I liked/didn't like the text length.
3. I liked/didn't like the options
4. The game mechanics is working/not working
and so on...
We didn't expect anyone to fall in love with the first proto. The feedback however came in a very different form:
1. Can I see the characters?
2. Would be great if there is movement in the characters
3. Can there be physical interactions between characters?
It was insanely interesting that the people who were claiming to want a Text Based RPG, when provided the same, were actually asking for proper RPG games with characters and animations.
Why, people engaged in building with AI for gaming, are going the Text Based RPG route?
Because we're used to finding linear solutions. GPT came first; we're most familiar with text; we have the most control over text, and we're used to trying to build the MVPs and find value there.
Why is it unlikely to be the best route?
1. Small actual market for text based RPG.
2. The reasons that the market likes text based games run deeper than spontaneous creativity of an LLM.
3. Even the ones who are using AI Dungeon's and Arcane are asking for them to be more interactive than just being cooler with Text.
I believe that games will work best when created with visuals, mechanics and much more. Text requires much higher precision than art. Generative AI models creating images and art are already getting better and better. We can arguably create better games that have visuals, animations, movements, shooting and interactions than Text Based RPGs because of the quality of image, video, audio and code generating models. This is why we'll continue to do that at Chaotix.