Hub
ML
Marcelo Lorenzetti
The Mathematics Inside AI
Hub Responsible AI Accountability Mathematics
SIMULATED DATA
Thought Leadership · Chapter 3 of 3

The Mathematics Inside AI

How language becomes mathematics — and how mathematics becomes behavior. A guided walkthrough of the transformations inside every modern AI system.

↓↓↓↓

Language is converted into numbers — embeddings that capture meaning.

Begin the Walkthrough Back to Hub
Live Demo · The Disciplines in Sequence

From ingestion to retrieval — the mathematical disciplines in sequence

Follow a corpus of documents through the full data lifecycle. Each stage is a live, interactive application of a mathematical discipline: tokenization, inferential statistics, linear algebra, calculus, probability, and descriptive geometry — culminating in retrieval.

Hover the dotted terms above to see a plain-English definition.

The Mathematics of Data: Ingestion → Retrieval
Watch the mathematical disciplines act in sequence as data is ingested, transformed, and retrieved.
Stage 1 / 7 · Tokenize
1TokenizeIngestion Interactive

Raw data arrives as documents. Before any math can act, the text is split into tokens — the atomic unit of every transformation that follows. Click a document to include or exclude it from the corpus.

Corpus
Tokenize refund policy
refund policy
Language is now a bag of tokens. Every discipline that follows operates on these tokens.
Interactive Walkthrough

From language to behavior, one transformation at a time

Each stage is a fully interactive step in the AI pipeline. Move through the walkthrough to see how words become numbers, numbers become predictions, and predictions become the behavior we observe.

Live Demo

The Mathematics Inside AI

AI is not magic — it is a sequence of mathematical transformations. Follow one sentence through the math.

SIMULATED DATA
2Tokens

Input sentence: “Governance makes AI safer.”

The model cannot read text — it must first split it into tokens and assign each an identifier.

"Govern"
5143
"ance"
872
" makes"
2304
" AI"
9821
" safer"
7412
"."
13
Language becomes numbers before the model can process it.
At the core of modern AI
LANGUAGE→TOKENS→NUMBERS→VECTORS→MATRICES→ATTENTION→PROBABILITY→SELECTION→OUTPUT

Language becomes mathematics — and mathematics becomes behavior.

Executive Summary

Three takeaways

The chapter, distilled for the reader in a hurry.

1Takeaway 1

Language becomes mathematics

Every AI system begins by converting text into tokens and vectors — meaning becomes geometry a computer can compute.

2Takeaway 2

Each discipline plays a distinct role

Statistics estimates, linear algebra represents, calculus optimizes, probability scores, geometry visualizes — the pipeline is a sequence of mathematical disciplines.

3Takeaway 3

Understanding the math enables trust

Knowing how predictions are made — and retrieved, audited, and explained — is the foundation the governance and accountability chapters build on.

Previous Chapter
Chapter 2
AI Liability & Accountability
Next Chapter
Chapter 1
Responsible AI as Competitive Advantage
The argument continues

Mathematics is the foundation — trust and accountability are the practice

Understanding how AI works internally is the beginning. Chapters 1 and 2 show how organizations turn that understanding into trustworthy, accountable practice.

Back to Thought Leadership Hub
Marcelo Lorenzetti
Author & Speaker · Guild Systems (professional affiliation)
Evergreen educational asset · Simulated demonstration data