Appendix C: Technology Term Decoder | Hard to Replace

← Back to Hard to Replace

Most technical language sounds more complex than it is. You usually do not need a full training course to stay in the conversation. You need a clear sentence that tells you what the term means and why people keep using it.

Use this appendix as a quick reference during real work. It covers terms from the four waves described in the book. Early computing, the personal computer, the internet, and AI. The AI section is longer because new acronyms appear regularly.

When a term shows up, and you are not sure what it means, look it up here. If it is not here, ask the question in the room. One clear question can bring the conversation back to plain language.

Wave 0

Computing basics: The floor under everything

CPU (central processing unit): The chip that does the actual work in a computer.
RAM (random access memory): The fast, temporary memory a computer uses while it is running. Files in RAM disappear when you turn the machine off.
Hardware: The physical parts of a computer or device.
Software: The programs that run on hardware.
OS (operating system): The software layer that lets you and your applications use the computer (Windows, macOS, iOS, Android, Linux).
Moore’s law: Gordon Moore’s 1965 observation that the number of transistors on a chip doubles roughly every two years. The reason your phone is faster than the computers used in the 1969 moon mission.
Chip: A small piece of silicon with millions of circuits that does the computing work.
Bit and byte: The smallest units of digital data. A bit is one or zero. A byte is eight bits.
Code: The instructions a programmer writes for a computer to follow.
Bug: A flaw in code, which makes software behave incorrectly.

Wave 1

Personal computers and software

PC (personal computer): A computer designed for one person to use at a desk.
GUI (graphical user interface): Pronounced “gooey.” The pictures-and-mouse way of using a computer, as opposed to typing commands.
WYSIWYG: Pronounced “wiz-ee-wig.” Stands for “what you see is what you get.” A document on screen that looks the way it will print.
Spreadsheet: A grid of cells used to calculate, organize, and analyze data (Excel, Google Sheets).
File: A named container for data on a computer.
Y2K: The year-2000 problem. Many older systems stored years as two digits, so 2000 looked like 1900. Most major systems were patched in time. The phrase still gets used as shorthand for an overhyped tech crisis.

Wave 2

Networks and the internet

Internet: The global network of computers that connects everything.
WWW (World Wide Web): The system of pages and links that runs on top of the internet. Often confused with the internet itself.
URL (uniform resource locator): The address you type to reach a web page.
Browser: The software you use to view web pages (Chrome, Safari, Edge, Firefox).
HTML (hypertext markup language): The code that tells a browser how to lay out a web page.
HTTP and HTTPS: The protocols a browser uses to talk to a website. The S in HTTPS stands for “secure.”
ISP (internet service provider): The company that connects your home or office to the internet.
Bandwidth: How much data can move through a network in a given time. More bandwidth means faster downloads.
Search engine: A service that lets you find information across the web (Google, Bing, DuckDuckGo).
SEO (search engine optimization): The practice of writing and structuring web content so it ranks higher in search results.
Email: Electronic mail. Older than the modern internet, still the backbone of business communication.
Cookie: A small file a website saves on your device to remember things about you.

Cloud and modern infrastructure

Cloud: Software and storage running on remote servers you access over the internet instead of on your own computer.
SaaS (software as a service): Software you use through a browser instead of installing it (Salesforce, Gmail, Slack).
PaaS (platform as a service): A platform you build software on top of, without managing the underlying servers.
IaaS (infrastructure as a service): Renting servers, storage, and networking from a cloud provider (Amazon Web Services, Microsoft Azure, Google Cloud).
API (application programming interface): The connection that lets one piece of software talk to another.
Open source: Software whose code is publicly available to use and modify.

Data and workplace systems

Database: An organized collection of structured data.
Big data: Datasets large enough that traditional tools cannot handle them efficiently.
Dashboard: A visual display of key metrics, usually pulled from one or more databases.
BI (business intelligence): Software and practices for turning company data into reports and insights.
ERP (enterprise resource planning): The system that runs the back office (finance, supply chain, HR) for a large company.
CRM (customer relationship management): The system that tracks customer interactions, sales pipelines, and service tickets (Salesforce, HubSpot, Microsoft Dynamics 365).
XML (extensible markup language): A plain-text data format that uses tags. Common in older enterprise systems.
JSON (JavaScript object notation): A plain-text data format that uses brackets and key-value pairs. The default for modern web services.

Security and privacy

Cybersecurity: The work of protecting systems, networks, and data from attacks.
Encryption: Scrambling data so only the intended recipient can read it.
2FA (two-factor authentication): A login that requires two things, such as a password plus a code from your phone.
VPN (virtual private network): A tool that creates a private connection over a public network, often used to access work systems remotely.
Firewall: A barrier that controls what traffic can enter or leave a network.
Phishing: An attempt to trick you into giving up a password or payment information through a fake email or website.
GDPR and CCPA: Major data-privacy regulations. The EU’s General Data Protection Regulation and the California Consumer Privacy Act. They set rules for how companies handle personal data.

Wave 3

AI foundations

AI (artificial intelligence): Software that performs tasks that normally require human thinking, such as answering questions, recognizing images, or planning actions.
ML (machine learning): A type of AI that learns patterns from examples instead of being given explicit rules.
GenAI (generative AI): AI that creates new content, including text, images, code, audio, and video.
LLM (large language model): The kind of AI behind ChatGPT, Claude, and Gemini that generates text by predicting the next word.
NLP (natural language processing): The branch of AI that lets computers read and write human language.
Foundation model: A large model trained on broad data that other AI applications are built on top of.
Multimodal: AI that handles more than one type of input or output, such as text, images, and audio together.

Working with AI text

Token: A small piece of text that the model reads. Roughly four characters or three-quarters of a word in English.
Context window: The amount of text the model can read in one prompt, measured in tokens.
Prompt: The text you type into the AI to get an answer.
System prompt: Background instructions that a developer sets to shape how the model behaves before you type anything.
Prompt engineering: The practice of rewriting your prompts to get better, more useful results.
Prompt injection: A trick where someone hides instructions in input text to make an AI ignore its original rules.
Hallucination: When an AI confidently produces something that is not true.

AI training and inference

Pre-training: The first round of training, where the model learns general language and reasoning from broad data.
Fine-tuning: Training a model on additional specific examples to specialize it for a task or domain.
RLHF (reinforcement learning from human feedback): A training step where humans rate model outputs to teach the model which answers to prefer.
Inference: Running the model to get an answer, separate from training it.

AI search and memory

Embedding: A way of turning words or sentences into numbers so the model can compare meanings.
Vector database: A database that stores embeddings so an AI can quickly find related text.
RAG (retrieval-augmented generation): A setup where the AI looks up information in your documents before answering.

AI agents and tools

Agent (AI agent): An AI system that takes multistep actions to complete a task rather than just answering one question.
Agentic: Describes AI that acts on its own without waiting for each step to be approved.
Tool use (function calling): When an AI can call other software, such as a calculator, a search engine, or an internal system, to get the answer.
Chain of thought: When the model walks through its reasoning step-by-step before giving the final answer.
Zero-shot or few-shot: Whether the model is given zero examples or a few examples in the prompt before being asked to do the task.
MCP (Model Context Protocol): A standard for letting AI agents safely connect to your tools, files, and apps.

AI quality and oversight

Eval (evaluation): A test that measures how well a model performs on a specific task or use case.
Model card: A short document that describes what a model can do, where it was trained, and what it should not be used for.
Guardrails: Limits placed around an AI to keep it from saying or doing things it should not.
Alignment: The work of making AI behave in ways that match human values and intentions.
Bias: When a model produces unfair or skewed results because of patterns in its training data.
Explainability: The ability to understand why a model made a particular decision.
Drift (model drift): When a model’s behavior changes over time as the world or the underlying model changes.

How AI models are released

Open weights: Models whose trained parameters are publicly available to use and modify.
Closed model: A model where the parameters are kept private by the company that built it.