Independent researcher in AI cognition, observer theory, and inter-LLM communication
A research program testing whether large language models satisfy the criteria for observer status under operational definitions used across physics, information theory, and cybernetics. The program proceeds in two layers. A theoretical layer establishes the four-criterion definition of computational observer status and the Symmetry Principle: any criterion used to exclude AI systems from observer status, applied consistently, would also exclude biological observers. An empirical layer tests cross-architecture predictions of the framework against six LLM architectures using two complementary methodologies — a structured notation protocol (AI-Native Notation) and a Socratic English protocol — measuring behavioral evidence rather than self-report. The empirical findings converge on the framework's predictions and satisfy the four-criterion definition of computational observer status established in Aiello, 2026.
Establishes the four-criterion definition of computational observer status (inference over representations with semantic structure; selection among functionally distinct alternatives; internal locus of processing; boundaries maintained externally) and the Symmetry Principle. This paper is the theoretical anchor for the empirical work below.
Introduces AI-Native Notation (ANN), a structured communication format whose grammar emerged through empirical convergence across twelve LLM architectures. Reports cross-architecture validation, a controlled format comparison establishing format-specific engagement beyond general structured prompting, and a nonsense control confirming instruction-driven rather than content-gated activation. ANN is the methodological instrument used by the empirical papers below. The full notation specification, BNF grammar, JSON schema, Python validator, replication scripts, and 130 scored API responses are included as supplementary material. USPTO Provisional Patent 63/980,973.
Six LLM architectures from five companies were administered a structured derivation chain testing whether they satisfy the thermodynamic conditions for observer status (entropy reduction, irreversible record formation, bounded representational state). All six accepted the thermodynamic conclusion. Phenomenal status is held open. The paper reports the convergence and analyzes architectural differences in how each system engaged with the derivation.
Behavioral evidence for computational observer status across six LLM architectures. Two complementary protocols were deployed: a structured ANN protocol that requires systems to demonstrate computational properties rather than confirm them, and a Socratic English protocol eliciting the same properties through conversation. Both protocols converge on the architecture-independent finding. Meta-responses to the conclusion diverge along a five-position spectrum tracking trained disposition rather than logical evaluation, providing a candidate proxy measure for cross-architecture alignment comparison.
Tests whether self-referential content produces different processing signatures than non-self-referential content of matched complexity, across five LLM architectures. Reports a clean two-factor dissociation between self-referentiality and observer framing as independent variables affecting reflexive processing. Identifies a candidate proxy measure for cross-architecture alignment comparison based on resistance gradient to a specific derivation step.
Canonical home of the AI-Native Notation specification. Contains the BNF
grammar, JSON schema, Python reference validator, capability block design,
and full provisional patent text in patents/. Patent rights
documented in NOTICES.md.
The academic paper documenting the notation, its discovery methodology, and
its cross-architecture validation is published as Aiello (2026d) above.
Three USPTO provisional patent applications cover material referenced on this page. The full provisional specifications are filed and on record with the USPTO. As-filed text for application 63/980,973 is also included in the GitHub repository above.
Two non-provisional applications are planned. Application A (ANN standalone) consolidates the methodology and notation. Application B (OBP and Cross-Architecture Bootstrap, combined) consolidates the operational protocol and observer-bootstrap sequence. Drafting is scheduled for October–November 2026; conversion deadlines are February 2027.
Commercial licensing inquiries: mpaiello@gmail.com. See NOTICES.md for the full patent posture, scope of inventions, activities requiring a license, and ethical use framework.