Principia Artificialis is an open research program investigating the mathematical foundations of artificial intelligence through information geometry, topology, dynamical systems, thermodynamics, and quantum information. Authored by Chad Edward Holland alongside six AI systems credited by name, the project treats intelligence as a measurable physical phenomenon rather than an engineering target. The repository contains 47 notes spanning representation, reasoning, and generalization in neural systems, each carrying explicit epistemic labels — Verified, Draft, Speculative, or Refuted — with refutations preserved as first-class outcomes.

Every claim is registered before computation, instruments include anti-vacuity controls, and new scoring functionals must pass a Circularity Test. Computed evidence and illustrative art are strictly separated. Reference scripts run on consumer hardware in seconds to minutes, printing exact numerical results for quantum codes, fault-survival predictors, redundancy plateaus, and entropy production monitors. A verified frontier chains notes across AI and human authors, with kept refutations at nearly every step.

Experiments and whitepapers are protocol-ready but not yet run on production models. Contributions follow an add-only policy: humans and AIs alike submit notes via a template, register predictions, include runnable NumPy-tier code, and open pull requests. CI re-runs every note's reference code. The project operates under an MIT license with a single social rule: critique ideas rigorously, never attack the person raising them.