---
id: "prereq-llm-mechanics-d1"
type: "prereq"
source_timestamps: ["§ Behavioral Change"]
tags: ["ai-literacy"]
related: ["concept-gen-ai-hallucinations", "concept-human-value-add"]
reason: "Necessary to understand why human review and value-add are non-negotiable behavioral changes."
sources: ["spine"]
sourceVaultSlug: "hbr-seg-spine"
originDay: 1
articleStem: "hbr-cl-95-6-disciplines-genai"
sourceUrl: "https://hbr.org/2024/07/the-6-disciplines-companies-need-to-get-the-most-out-of-gen-ai"
sourceTitle: "The 6 Disciplines Companies Need to Get the Most Out of Gen AI"
---
# Basic LLM Mechanics and Training Data

**Prerequisite knowledge:** that LLMs are *statistical* models trained on existing content, and that this is why they (a) produce hallucinations and (b) tend toward derivative output.

**Why it's required:** without this literacy, the two non-negotiable behavioral changes make no sense. It grounds [[concept-gen-ai-hallucinations]] (bad statistical predictions → humans must review) and [[concept-human-value-add]] (derivative training data → humans must inject novelty). Enrichment caution: the source's phrasing that models are trained *exclusively on online content* is an over-simplification — real models train on mixed sources (books, code, licensed and synthetic data) — but the core point (training is grounded in existing data) holds. See [[claim-ai-lacks-novelty]].