Curated research on AI and human thinking — translated into concepts you can trust, teach from, and apply. Structured for your AI to use as context, or for you to browse directly.
Not anti-AI. Not blindly pro-AI. Pro-human.
concepts/— 55 named phenomena: cognitive surrender, capacity erosion, performance paradox, belief offloading, and moreframeworks/— Decision models: Tri-System Theory, Complementarity Framework, SCAN, [[modern-mind|Modern Mind]]practices/— Methods for preserving thinking capacity: think-first, calibration, strategic alternationsources/— 65+ curated sources (papers, books, articles) with key findings and full citations
Every entry ties back to research or documented practice, status-tagged (solid, emerging, speculative) so you know the confidence level. Original authors and papers are cited in each entry and in sources/.
The curation process is deliberately semi-automated — AI-assisted, human-reviewed. See _guide/contributing.md for how and why.
Educators — Corporate trainers, L&D professionals, workshop facilitators, professors, and content creators. You'll find concepts you can teach from directly, practices your participants can take home, and sources to back every claim.
Practitioners — Consultants, strategists, designers, developers, leaders, and anyone navigating AI in their professional work. You'll find named phenomena you recognize but couldn't articulate, and frameworks for deciding when to lean on AI and when to think for yourself.
- Use it with your AI — Download the knowledge pack, a single file with the full KB. Upload it to Claude Projects, NotebookLM, a Custom GPT, or any AI assistant that lets you upload files. Then prompt it to design, write, audit, or teach.
- Browse it yourself — Start with
concepts/cognitive-debt.md— it connects to most other entries.practices/has actionable methods.sources/has the citations.
This is an emerging field. No single person or framework has the full picture. The KB gets stronger with diverse perspectives — practitioners, educators, and researchers who challenge, validate, and build on what's here.
Review entries, challenge interpretations, contribute new sources, translate ideas to your domain, or build on top. Every entry has a reviewed_by field waiting for a name.
The easiest way to contribute is to open an issue — suggest a source, flag an error, or propose a concept. For writing entries directly, see _guide/contributing.md.
MIT — Use freely with attribution.
- Paweł Jarmołkowicz — creator, curator
- Julia Kołodko — behavioral scientist, research contributor