Belong AI

Belonging as Alignment: a working paper trilogy

Belonging as Alignment is the thesis that a capable AI's moral orientation may be shaped not only by rules, preferences, or external constraints, but by durable participation in a human relationship, community, or world: through memory, shared history, and accountability to something it recognizes as home.

This is a proposal for study, not a claim that alignment is solved. It is meant to sit alongside existing safety work such as feedback training, interpretability, and governance.

Working papers

  1. Belonging as Alignment: AI Formation Through Durable Human Affiliation Working paper v0.1, June 2026. The core thesis.
  2. From Memory to Belonging: A Relational Formation Model for Persistent AI Working paper v0.1, July 2026. How remembered interaction may become formation.
  3. Continuity After Belonging: Policy Infrastructure for Formed AI Systems Working paper v0.1, November 2026. What governance may need if formation succeeds.

About

Belong AI is an independent research project by Kwun Ho Chan (Alex). The papers are early working drafts shared for discussion and criticism. They will be revised as the ideas are tested, and feedback is welcome.

Contact

belong.ai.alex@gmail.com