HOW WE WORK
We research 3,627 topics across 34 disciplines. Every claim is rated by evidence quality. Every source is verified against academic databases. This section explains how — and doesn't hide the limitations.
THE FULL PICTURE — five sections, no spin
How We Rate Every Claim
The 4-tier evidence system (Verified → Dubious), [N/5] source confidence scoring, interpretive lenses, and the non-negotiable rules applied to every document in the corpus.
→What Gets Verified and How
External API verification (CrossRef, Retraction Watch, Open Library, PubMed), the 8-dimension quality pipeline, and what "automatically checked" actually means in practice.
→Current State of the Corpus
Live quality metrics for all 3,627 documents — scored on 8 quality dimensions and 7 factuality dimensions. These numbers update as the corpus is improved.
→How AI Is — and Isn't — Used
A complete disclosure of what the AI partner does, what it doesn't do, what the adversarial review program found, and how we've built institutional checks against AI overconfidence.
→Is This AI Garbage? And Other Honest Questions
The questions you're actually thinking — from hallucinated citations to conspiracy-theory concerns to whether you can trust AI-researched content. Answered without spin.
→Convinced? Skeptical? Either is fine. Start with the research base itself — every document is open, every source is listed.
Explore the Research Base →