Over the weekend, we quietly rolled out something new that’s pretty cool (at least we think so!): RLHF controls.
If “RLHF” sounds like a new texting slang, it actually stands for Reinforcement Learning with Human Feedback. Simply put, it’s a fancy way of saying that you can now teach our AI model to get smarter about Dramatica theory—just by clicking a thumbs-up or thumbs-down.
So, here’s the ask:
If you’re a Dramatica pro (or at least know your Universe from your Physics), we’d love your help fine-tuning our AI. All you need to do is:
- Go to your Preferences and flip on the AI Training option.
- Whenever the AI gives you a response that’s a bit “off,” click thumbs-down and quickly let us know what it got wrong.
- If it nails it, thumbs-up helps too!
This feedback directly improves how accurately the AI understands and explains Dramatica theory. And don’t worry, you don’t need to write a dissertation—a brief note on why the answer was off is plenty.
Think of it as your chance to shape the narrative (literally).
Thanks for helping us make Subtxt/Dramatica even better!