Agustin V. Startari's Blog, page 2

How One Bad Sentence Can Cost Your Company Millions: When Grammar Fails the Audit

AI-powered systems are misclassifying corporate expenses, not because they lack data, but because they misread grammar. What looks like a...

How Generative Models Misclassify Business Transactions, and Why It Is a Structural, Not Semantic, Problem

1. What the Article Explores The forthcoming paper Expense Coding Syntax: Misclassification in AI-Powered Corporate ERPs  addresses a...

When Grammar Sells You a Lie: How AI Whitepapers Are Structurally Built to Deceive

This article investigates how AI-generated crypto whitepapers use persuasive grammar to simulate financial credibility. Through analysis of 10,000 documents and a custom syntactic risk model (DSAD), it demonstrates that sentence structure—not content—is a key predictor of project failure. The paper proposes a regulatory framework called fair-syntax governance to audit and reduce this emerging risk.

How a 12% Tax Rule Becomes Code: Grammar, Law, and Machine Execution

Grammar, Law, and Machine Execution What This Article Explains In the shift toward automated legal enforcement, one key transformation is...

Think Your AI Understands You? It Already Started Responding

AI systems often act before any real understanding takes place. This is not a design footnote; it is a structural risk in law, medicine,...

What If Your AI Was Already Acting Before You Spoke?

AI or not AI, that is the question What this article is about ( AI) In Pre-Verbal Command: Syntactic Precedence in LLMs Before Semantic...

Law Doesn’t See the Code

Why Legal Systems Fail Against Executable Sovereigns When responsibility disappears and rules execute themselves, the law is left chasing...

Synthetic Ethos: When Credibility Is Coded Without Source

Large language models now simulate credible voices without authors or references. This engineered “synthetic ethos” poses epistemic risks in domains like healthcare, law, and education. Based on 1,500 AI-generated texts, the article shows that persuasive fluency is prioritized over traceability, requiring structural responses to detect and regulate credibility without source.

Grammar Without Judgment: How One Rule Erases Ethics from AI Execution

This article demonstrates how ethical judgment can be structurally erased from AI grammars using the deletion rule δ:[E] → ∅. It shows that moral content can be removed at the syntactic level without semantic suppression or loss of computational validity.

Executable Power Is Not a Metaphor. It’s Code.

Introduces executable power as subjectless authority based on syntactic rules. Empirical analysis of LLMs, TAPs, and smart contracts shows 100% execution under formal triggers. Marks a break from narrative models of power toward structural operativity.