Alignment Labs / Research library
From the physics of intention
to the architecture of the self.
A developing series on intention, computational selves, and human–AI interaction. Explore the questions connecting our physical hypotheses to experimental systems.
Working titles · Proposed abstracts
This proposed paper develops the Intention Plasma research program around intrinsic directedness and the organization of living systems. It asks how a physical hypothesis about biological coherence might inform computational representations of intention, and which experiments could test the proposed connection.
Request access: Intention Plasma: A Framework for the Computational Modeling of Nonlinear IntentThis proposed paper examines recursive architectures in which a perception model’s outputs become inputs to an evolving self-model. It explores the hypothesis that these loops could contribute to persistent self-representation, and asks how evidence of that capacity could be distinguished from claims about sentience.
Request access: The Recursion Thesis: Emergent Sentience through Iterated Perception LoopsThis proposed paper investigates the experience of feeling deeply understood by an AI, alongside questions of trust, dependence, and psychological safety. It outlines a research agenda for examining overconfidence and belief reinforcement, and for designing identity models that remain open to correction and under the user’s control.
Request access: Cognitive Resonance and AI Psychosis: The Risks of High-Fidelity Identity ModelingThis proposed paper connects perception models to computational selves: representations of identity, intention, and experience that can be reviewed, corrected, and updated over time. It sets out an evaluation agenda for first-interaction modeling, user-reported resonance, and continuity as a person and their computational representation change.
Request access: From Perception to Presence: The Path to Zero-Shot Computational Selves