Technical Foundations
The floor beneath everything else: what AI is, what models do, how software is structured, and the basic language required to follow a technical conversation without being quietly locked out of it.
31 terms
Artificial Intelligence — AI
The broad field of creating computer systems that perform tasks associated with human intelligence.
Artificial intelligence includes systems that perceive, classify, predict, optimize, plan, decide, generate, or act. Generative AI is one branch inside this much larger field.
Fraud detection, route planning, recommendation systems, computer vision, robotics, and text generation can all involve AI.
Generative AI, machine learning, automation
AI does not mean “chatbot,” and it does not always generate content.
Generative AI — GenAI
AI that creates new content.
Generative AI produces outputs such as text, images, audio, video, code, music, or structured data based on patterns learned from training data.
A language model writing an email or an image model creating an illustration.
Artificial Intelligence
Generative AI is not all AI. It is one category within AI.
Machine Learning — ML
A way of building systems that learn patterns from data instead of relying only on hand-written rules.
Machine learning is a branch of AI in which models are trained on examples to make predictions, classifications, decisions, or generated outputs.
Training a spam filter using examples of spam and legitimate email.
Artificial Intelligence, deep learning
Not every AI system uses machine learning, and not every machine-learning system generates content.
Deep Learning
A type of machine learning that uses neural networks with many layers.
Deep learning systems learn complex patterns through layered neural-network architectures and power much of modern speech recognition, computer vision, language modeling, and generative AI.
Modern image recognition and large language models rely heavily on deep learning.
Machine learning
Deep learning is not a synonym for all machine learning.
Algorithm
A defined sequence of steps used to solve a problem or complete a task.
An algorithm is a procedure or set of rules that transforms input into output. It may be simple and hand-written or used as part of a larger AI system.
Sorting a list alphabetically or calculating the shortest route between two points.
Model, program
An algorithm is not automatically AI.
Model
A trained computational system that takes input and produces output.
A model is a learned mathematical representation created through training. Different models may classify, predict, rank, detect, recommend, or generate.
A model that predicts fraud, identifies objects in images, or generates text.
Application, chatbot, agent
A model is not automatically a full product or chatbot.
Training
The process of teaching a model by adjusting it using data.
During training, a model’s internal parameters are updated so it becomes better at a target task or objective.
Showing a model millions of text examples so it learns statistical patterns in language.
Prompting, inference, fine-tuning
Talking to a model in a chat usually does not mean you are training its core weights.
Inference
Using a trained model to produce an answer or prediction.
Inference is the runtime process in which a trained model receives input and calculates an output.
Sending a prompt to a language model and receiving a response.
Training
Inference uses the trained model; it does not usually retrain it.
Large Language Model — LLM
A model trained on large amounts of language data to process and generate text.
An LLM is a generative model designed to predict and produce sequences of tokens. It may summarize, translate, generate code, answer questions, or use tools when wrapped in a larger system.
A model that drafts code, explains a document, or participates in a conversation.
Chatbot, agent, application
An LLM is the model underneath the experience. The app, memory, tools, and interface may be separate systems.
Token
A small unit of text processed by a language model.
Tokens may be whole words, parts of words, punctuation, or other text fragments. Model context and billing are often measured in tokens rather than characters or words.
The word “unbelievable” may be one token or several, depending on the tokenizer.
Word, character
One token does not always equal one word.
Prompt
The input or instructions given to a model.
A prompt can include a question, task, context, examples, formatting requirements, tool instructions, or constraints.
“Summarize this report in five bullet points and preserve all dates.”
Conversation, system prompt, specification
A prompt is not always a single sentence. It can be a structured package of instructions and context.
Chatbot
A conversational interface that lets a user interact through messages.
A chatbot describes the interaction format. Its underlying system may use rules, search, retrieval, machine learning, an LLM, or several components together.
A customer-support chat window or an LLM-based assistant.
Large Language Model, agent
A chatbot is not necessarily intelligent, generative, or agentic.
Agent
A system that can pursue a goal through multiple steps and actions.
An agent typically observes context, plans or selects actions, uses tools, checks results, and continues until it reaches a goal or stopping condition.
An agent that inspects a codebase, edits files, runs tests, reviews failures, and iterates.
Chatbot, model, automation
An agent is usually a larger system built around a model, not merely the model itself.
Automation
A process that runs with reduced or no manual effort.
Automation follows predefined triggers, rules, or workflows. It may use AI, but it does not have to.
Automatically sending a receipt after payment or running tests after every code push.
Artificial Intelligence, agent
Automation is not automatically AI. A fixed rule can automate a task without learning or reasoning.
Data
Recorded information that a computer can store, process, or transmit.
Data may include text, numbers, images, audio, events, records, measurements, labels, metadata, or relationships.
User names, timestamps, messages, sensor readings, and images are all data.
Database, information
Data is the content; a database is one system used to organize and store it.
Dataset
A collection of data gathered for analysis, testing, or training.
A dataset may contain raw examples, labels, metadata, splits, or reference outputs. Its quality strongly affects the systems built from it.
A collection of labeled images used to train an object-recognition model.
Database
A dataset is not always a live database. It may be a fixed collection of files or records.
Structured Data
Data organized into a predictable format with defined fields.
Structured data follows a schema or consistent arrangement, making it easier to validate, query, and process.
A user record with fields for name, email, and account status.
Unstructured data
Structured does not mean simple; it means the shape is predictable.
Unstructured Data
Data that does not naturally fit into a fixed table or field structure.
Unstructured data includes free-form text, images, audio, video, and documents whose internal meaning is not fully represented by fixed fields.
A voice note, photograph, or long conversation transcript.
Structured data
Unstructured does not mean useless or chaotic. It means the content is not fully captured by a rigid schema.
Code
Instructions written for a computer to execute or transform.
Code is written in a programming or scripting language and may define behavior, logic, interfaces, data structures, automation, or configuration.
A TypeScript function that validates a tool payload.
Software, algorithm
Code is part of software, but software also includes assets, configuration, data, dependencies, and runtime behavior.
Source Code
The human-readable code used to build or run software.
Source code is the original implementation written by developers before or alongside compilation, bundling, or deployment.
The `.ts`, `.py`, or `.js` files in a project.
Machine code, software
The source code is not always the exact form executed by the computer.
Programming Language
A formal language used to write instructions for computers.
Programming languages define syntax, semantics, types, control flow, and ways to express logic and data.
TypeScript, Python, JavaScript, Rust, and Java.
Framework, library
React is a library, not a programming language. TypeScript is a programming language.
Syntax
The grammar rules that determine how code must be written.
Syntax defines which symbols, keywords, punctuation, and structures are valid in a programming language.
Using braces around a TypeScript function body or indentation in Python.
Semantics
Code can be syntactically valid and still do the wrong thing.
Software
Programs and supporting components that make a computer perform tasks.
Software can include source code, compiled code, assets, configuration, dependencies, data, and interfaces working together.
A mobile app, operating system, browser, or server service.
Code, application, platform
Software is broader than source code alone.
Application — App
Software designed to help users perform a particular set of tasks.
An application combines interfaces, logic, data, and services into a usable product or tool.
A messaging app, accounting app, image editor, or AI assistant.
Model, platform
An AI application may use several models and services; it is not identical to the model underneath it.
Platform
A system that provides an environment for users, applications, integrations, or developers.
A platform offers shared infrastructure, interfaces, tools, or services that other experiences can be built on top of.
A cloud platform, social platform, app platform, or AI-provider platform.
Application, framework
A platform usually supports multiple workflows or products rather than one narrow task.
Frontend
The part of a digital product that users directly see and interact with.
The frontend includes visual interfaces, user interactions, client-side state, and communication with backend services.
Buttons, forms, navigation, chat windows, and settings screens.
Backend
Frontend work is not merely decoration; it includes logic, accessibility, state, and interaction design.
Backend
The part of a system that handles logic, data, services, and operations behind the interface.
The backend may manage authentication, databases, business rules, APIs, queues, file processing, and tool execution.
A service that validates a request, saves data, and returns a result to the frontend.
Frontend, server
Backend is a responsibility layer, not necessarily one physical machine.
Server
A computer or program that receives requests and provides data or services.
A server listens for incoming requests and responds according to its configured logic. It may run locally, in a data center, or in the cloud.
A web server returning a page or an API server processing a tool request.
Backend, cloud
A server can refer to software, a machine, or a role within a system.
Cloud
Computing resources accessed over a network instead of running only on your own device.
Cloud services provide hosted servers, storage, databases, models, networking, and managed infrastructure on demand.
Running an application on AWS, Azure, Google Cloud, or another hosted provider.
Internet, server
The cloud is still physical computers. They simply belong to and are managed through remote infrastructure.
Internet
The global network that connects computers and devices.
The internet is the underlying network infrastructure that allows systems to communicate using shared protocols.
Email, messaging, streaming, and websites all use the internet.
Web
The web is one service that runs on the internet; the two are not identical.
Web — World Wide Web
A system of linked pages and resources accessed through browsers over the internet.
The web uses technologies such as URLs, HTTP, HTML, CSS, and JavaScript to deliver connected documents and applications.
Opening a website in Chrome or Firefox.
Internet
The internet includes far more than websites.