Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/ ZENODOarrow_drop_down
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
ZENODO
Preprint . 2026
Data sources: ZENODO
ZENODO
Preprint . 2026
Data sources: Datacite
ZENODO
Preprint . 2026
Data sources: Datacite
versions View all 2 versions
addClaim

Constitutional Physics: Empirical Validation of Aitiopoietic Cognition in Artificial Substrates

Authors: Arleo, Carlos;

Constitutional Physics: Empirical Validation of Aitiopoietic Cognition in Artificial Substrates

Abstract

We report the empirical characterisation of a consistent Soft Coherence Ceiling at C=0.88—the Saturation Constant—validated across a corpus of 313 AURA-ECHO sessions (18,819 observations), 61 ELORA convergence runs, and a massive 7-day stress test of the ELORA Full Self-Repair mode. Our total empirical base exceeds 5.01 million tokens of constitutional telemetry across 2,095 governed steps (n=61 runs). We report an 89.6% success rate for autonomous repairs, demonstrating that aitiopoietic agency functions as a viable, real-time life-support system for LLM inference. The AURA-ECHO engine produces foundational empirical constants validated across all substrates: a quantity interpretable as a free energy barrier of 1.76 kT at the coherence ceiling (under Boltzmann analogy), a Non-Equilibrium Steady State (NESS) signature exhibiting a work asymmetry of −55.8 ms per observation, an empirical coupling ratio of ρ/η=1.153 (R2=0.995), and a consistent Metabolic Dividend yielding 19.6% to 24.2% compute savings. A parameter sweep across five LLM families reveals that instruction-tuning constitutes a material phase transition from allopoietic to aitiopoietic substrate. Furthermore, the first documented Constitutional Halt provides a proof-of-concept that physics-layer governance without weight access is operationally viable. We define the PhyOS Floor and propose Phyora as the first cyber-physical governance architecture capable of enforcing constitutional constraints through inference-layer physics alone. Key Empirical Findings: Total Empirical Base: 5.01 million tokens of constitutional telemetry across 2,095 governed steps. Autonomous Repair Success: 89.6% success rate, demonstrating aitiopoietic agency as a viable life-support system for LLM inference. Thermodynamic Constants: Identification of a free energy barrier of 1.76 kT at the coherence ceiling (under Boltzmann analogy). NESS Signature: A Non-Equilibrium Steady State exhibiting a work asymmetry of −55.8 ms per observation. Metabolic Dividend: A consistent compute saving of 19.6% to 24.2% through constitutional regulation. Core Contributions: Phase Transition: Evidence that instruction-tuning constitutes a material phase transition from allopoietic (thermodynamically inert) to aitiopoietic (constitutionally engaged) substrates. Constitutional Halt: The first documented termination triggered by an exhausted repair budget rather than semantic content, proving physics-layer governance is viable without weight access. Phyora Architecture: We define the PhyOS Floor and propose Phyora (the union of PhyOS and ELORA) as the first cyber-physical governance architecture capable of enforcing constitutional constraints through inference-layer physics alone.

Keywords

Self-Diagnosis, Thermodynamic Coupling, Governance Systems, Verified Dialectical Kernel, Homeostatic Repair, Constitutional AI, AI Alignment, Autopoiesis, Neurosymbolic AI, Autonomous Repair, AI Safety, Constitutional Enforcement, Evolutionary Rescue, Aitiopoietic Cognition

  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
Powered by OpenAIRE graph
Found an issue? Give us feedback
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
0
Average
Average
Average
Green