AI Fundamentals for Governance Professionals Cheat Sheet

Updated 2026-07-31

Next Topic: AI Governance and Risk Management Cheat Sheet

🧠Study flashcards on this topic63 cards · spaced repetition→

AI Fundamentals for Governance Professionals covers the conceptual building blocks behind artificial intelligence: the different families of AI systems, how models actually learn, why they resist inspection from the outside, and the everyday vocabulary practitioners use when building and deploying them. This literacy matters because governance, compliance, legal and risk professionals are increasingly asked to assess, approve or challenge AI systems they did not build themselves, and a rule-based tool, a trained classifier and a generative model fail in fundamentally different ways that no policy language alone can paper over. The single most useful mental model here is the autonomy spectrum: nearly every governance question, from how much testing a system needs to how much a human must review before it acts, ultimately reduces to how much the system is allowed to do without a person checking its work. This sheet stays deliberately non-mathematical and does not cover specific laws or regulatory frameworks, which live in their own dedicated resources, so it can serve as the shared vocabulary underneath all of them.

What This Cheat Sheet Covers

This topic spans 10 focused tables and 63 indexed concepts. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.

Table 1: The AI Family Tree — Core System TypesTable 2: Foundation Models and the Generative AI WaveTable 3: How Machines Learn — Training ParadigmsTable 4: Training Data and the Model LifecycleTable 5: Under the Hood — Neural Networks in Plain TermsTable 6: The Black Box Problem — Opacity, Interpretability and ExplainabilityTable 7: Known Limitations and Failure ModesTable 8: The Autonomy Spectrum — From Human-in-the-Loop to Fully AutonomousTable 9: AI Agents and Agentic SystemsTable 10: Everyday Vocabulary for Working With AI

Quick IndexSubscribe to unlock

A jump-to index of every table row in this cheat sheet.

Mind MapSubscribe to unlock

An interactive map of every table and concept in this topic.

Table 1: The AI Family Tree — Core System Types

Every AI system a governance professional will encounter falls into one of a handful of broad families, and knowing which one you're looking at is the first step in any assessment. These categories nest inside one another, from the oldest and simplest (explicit rules) to the newest and least predictable (generative models built on foundation models).

System Type	Example	Description
Rule-based systems (expert systems, symbolic AI)	`IF patient has fever AND cough AND difficulty breathing THEN flag possible pneumonia` (a MYCIN-style diagnostic rule)	Behavior comes from explicit if-then logic written by human experts, not learned from data. • Every decision traces back to the specific rule that fired, making it highly auditable • Cannot handle situations its rules don't cover.
Machine learning (ML)	A product-recommendation model learns from a shopper's past purchases rather than following a fixed catalog of rules	The system learns statistical patterns from historical data instead of following hand-coded logic; machine learning is a subset of AI, and its accuracy generally improves as more data is fed in.
Deep learning	A neural network with dozens of stacked layers reads raw pixels to sort photos into "pizza," "burger" or "taco" without a human first specifying which visual features to look for	A subset of machine learning that uses multi-layered ("deep") neural networks to automatically discover relevant features in raw, unstructured data such as images or text, instead of relying on human-engineered features.
Generative AI	A user asks a chatbot to draft a one-page policy memo and receives new, original text rather than a retrieved document	A subset of deep learning, typically built on foundation models, that creates new content — text, images, audio, code — rather than only classifying or predicting on existing data.

Table 1: The AI Family Tree — Core System Types

System Type	Example	Description
Rule-based systems (expert systems, symbolic AI)	`IF patient has fever AND cough AND difficulty breathing THEN flag possible pneumonia` (a MYCIN-style diagnostic rule)	Behavior comes from explicit if-then logic written by human experts, not learned from data. • Every decision traces back to the specific rule that fired, making it highly auditable • Cannot handle situations its rules don't cover.
Machine learning (ML)	A product-recommendation model learns from a shopper's past purchases rather than following a fixed catalog of rules	The system learns statistical patterns from historical data instead of following hand-coded logic; machine learning is a subset of AI, and its accuracy generally improves as more data is fed in.
Deep learning	A neural network with dozens of stacked layers reads raw pixels to sort photos into "pizza," "burger" or "taco" without a human first specifying which visual features to look for	A subset of machine learning that uses multi-layered ("deep") neural networks to automatically discover relevant features in raw, unstructured data such as images or text, instead of relying on human-engineered features.
Generative AI	A user asks a chatbot to draft a one-page policy memo and receives new, original text rather than a retrieved document	A subset of deep learning, typically built on foundation models, that creates new content — text, images, audio, code — rather than only classifying or predicting on existing data.