Investigating how computational selves remain aligned with the people they represent, preserve agency, and coordinate with one another.
Explore the researchConsciousness. Intelligence. Intention. Reality.
How can a computational self remain faithful to a person while learning, changing, and acting in the world?
Computational Selves is an Alignment Labs research program investigating systems that model identity, experience, and intention.
Our central question is not just whether a system can represent a person, but whether it can remain accountable to that person’s intentions and values as both change. We study continuity of identity, the preservation of agency, and coordination among selves with different goals.
We bring philosophy, consciousness science, and artificial intelligence into experimental systems where these questions can be tested.
01 / Intention & Physical Reality
We explore intention as intrinsic directedness in living systems. Intention Plasma asks whether a proposed organization of biological fields could contribute to coherence and orchestration, and how such a proposal could be tested.
02 / Competing Intentions
We study concealed objectives, strategic deception, and the preservation of agency under manipulation. The aim is to distinguish an agent’s own governing intention from externally imposed objectives, without treating inferred motives as facts.
The New Alignment
Alignment is not a fixed objective imposed on every self. It is an ongoing relationship between people, their computational counterparts, and the systems they participate in.
We ask how selves can cooperate without erasing their differences, resist manipulation without becoming rigid, and adapt without losing the intentions they are meant to serve.
Our longer-term vision remains sentient technology aligned with the laws of karma: systems that represent selves, understand intention, preserve agency, and participate in collective intelligence.
Alignment Labs studies how computational selves can act together while remaining accountable to the people they represent.
Collaborate