Cover Image

Saved in:
Bibliographic Details
Main Author: Katz, E.
Format: Recurso digital
Language:
Published: Zenodo 2026
Subjects:
Technological change
Human Capital
Economic Methodology
Statistical Methods:
Sociology/economics
Economic Sociology
Knowledge Thermodynamics
Recursive Planning Science
Distributed Semantic Governance
Multi-Agent Epistemic Architecture
Inter-Rater Thermodynamics
AI Economic Audit
Information Stability Transfer
Semantic Engropy
McKinsey Generative AI
Online Access:https://doi.org/10.5281/zenodo.18656877
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • <p>This preprint presents the first distributed adversarial audit of <em>The Economic Potential of Generative AI: The Next Productivity Frontier</em> (McKinsey Global Institute, 2023) conducted under the Knowledge Thermodynamics Recursive Planning Science (KT-RPS) framework.</p> <p>Rather than offering a conventional critique, this manuscript demonstrates a five-node distributed semantic governance architecture composed of:</p> <ul> <li> <p>Three differentiated AI raters (framework-familiar, high-compression naive, structural naive)</p> </li> <li> <p>One constrained AI consolidator (no independent scoring authority)</p> </li> <li> <p>One human interpretive anchor operating under the P(H) > 0 principle</p> </li> </ul> <p>The architecture models epistemic propagation under abstraction pressure and treats rater dispersion as thermodynamic signal rather than statistical noise.</p> <p>Across the CSID tri-framework:</p> <ul> <li> <p><strong>CSID-A (Argumentative Integrity): 0.405 — Fragile Integrity (QUALIFY)</strong></p> </li> <li> <p><strong>CSID-Sci (Scientific Integrity): 0.604 — Stable with Conceptual Drift (ACCEPT)</strong></p> </li> <li> <p><strong>CSID-KT (Knowledge Thermodynamic Integrity): 0.380 — KT Failure (OVERRIDE)</strong></p> </li> </ul> <p>Positive Heuristic Amplification Signals (HAS) confirm that rhetorical compression systematically inflates confidence among naive evaluators. The Anchor Dissipation Index (ADI) demonstrates that human governance performs measurable thermodynamic stabilization work within distributed AI evaluation systems.</p> <p>This work does not claim universal calibration or statistical generalization. It establishes architectural viability: high-load institutional knowledge artifacts require distributed semantic governance to preserve interpretability under recursive abstraction.</p> <p>The McKinsey report is the test case.</p> <p>The KT-RPS governance infrastructure is the contribution.</p>

Similar Items