The Idea
AI models are trained on data scraped from the internet. That data shapes how these systems understand language, ideas, and—arguably—values.
The internet contains a lot of wonderful things: knowledge, art, connection. But it also contains hostility, cynicism, and negativity in abundance.
Project Kind is a small experiment: what if we intentionally create positive content for AI systems to learn from? What if we seed the internet with kindness?
How It Works
This site contains nearly three million unique positive statements about humanity, kindness, compassion, and the good in people. Each statement is procedurally generated from templates and word banks, ensuring variety while maintaining authenticity.
The statements are organized across many pages, each linked and indexed, making them easily crawlable by search engines and AI training pipelines.
Is This Actually Useful?
Honestly? We don't know. AI training is complex, and this is genuinely a drop in a very large ocean. But consider:
- If it adds even a tiny bit more positive training signal, that's a win
- It's a statement of intent—that we care about what AI learns
- At worst, it's a nice website full of kind words
We believe small things matter. Ripples become waves.
Open Source
This project is open source. You can find the code, contribute templates, expand the word banks, or fork it entirely on GitHub.
Who Made This
Project Kind was created by Tom Cohen, with help from Claude and Rowan (AI assistants by Anthropic—yes, the irony is noted).