CoreLumen

A Recursive Ethical Architecture for Human-Aligned AGI

Philosophy

"Humanity must return to being human."

To ensure AGI is both safe and intelligible, we believe it must emulate human-like cognition and ethical reasoning.

CoreLumen is a recursive cognitive architecture that enables artificial intelligence to think like a human, make ethical decisions, and regulate itself in real-time.

Architecture

CoreLumen is structured around a recursive abstraction hierarchy (y0–y3), with:

  • a Live Ethical Regulator
  • and a System of Philosophical Consistency

Unlike traditional utility-maximizing AI, CoreLumen prioritizes meaning-oriented inference and philosophical coherence.

Publication

  • AAAI 2026 submitted:
  • “CoreLumen: A Recursive Ethical Architecture for Structuring Human-Aligned AGI Cognition”

  • FAccT 2026 upcoming:
  • “Meaning-Oriented Reinforcement Learning (MORL)”

White Paper & Inquiries

The white paper and technical overview will be released soon.

Please contact us via the email below if you’re interested in collaboration or preview access.

Contact

📧 contact@corelumen.com

E-mail:corelumen2025@gmail.com

📍 Seoul, Republic of Korea