I open this notebook in 2026 with a simple premise: what I learn about AI, cognition, and the alignment problem deserves a place where origin, context, and confidence are visible — not a chronological feed of opinions.
What lives here
- Gnoses (
/docs) — canonical units of knowledge. Each withmaturity,confidence,sources,backlinks. They evolve over time. Some start as drafts and mature; others are born canonical because they've survived years of exposure. - Notes (
/blog, this place) — links, quick reactions to readings, clippings of what's moving. Chronological. Short. Not meant to be final.
Why now
I'm halfway through an MSc in AI at UNICAMP, coordinate the Mensa Brasil AI group, lead Synerise in Brazil, and I'm building Gnostikon — a canonical knowledge layer so intelligent systems can stop lying about what they know. The notebook externalizes the research that feeds all of that.
What to expect
- Bilingual PT/EN — the definitive version of each canonical gnosis will eventually exist in both languages. Notes may stay only in PT (working language) or only in EN (reading language).
- No hype. No lorem ipsum. No invented numbers.
- Everything here is refutable. If you disagree with argument, open an issue on the repo or email me.
Next canonical gnoses: the alignment problem as a stoic search for truth,
cognition as a recursive world-model, language as hypothesis compression,
and why math predicts. Landing in he-004.
Welcome.