What Is Artificial Intelligence? Generative AI Explained Simply | Free AI Course: Practical Artificial Intelligence for Work

What you'll learn

Understand what artificial intelligence and generative AI actually are.
Learn the terms you'll hear at work and the main categories of AI tools.
Know what generative AI does well and where it fails.
See why learning AI now pays off in almost any job.
Run your first three tests with copy-and-paste prompts and avoid rookie mistakes.

In just a few years, artificial intelligence (AI) went from research labs to everybody's phone. Today it drafts emails, summarizes meetings, analyzes spreadsheets, creates images and writes code. This course isn't for engineers. It's for anyone who wants to use AI well at work, understanding just enough about how it works to get real value from it and avoid its mistakes.

What AI is
Key concepts
AI you already use
The tools
Your first tests

What artificial intelligence is

Artificial intelligence is the branch of computer science that builds systems able to do tasks that normally require human intelligence: understanding language, recognizing images, making decisions, translating or creating content.

The key difference from a traditional program looks like this:

Traditional programA person writes the rules

How it works: Exact rules: "If an email contains 'make money fast,' mark it as spam."
Strength: Predictable: same input, same output, every time.
Limit: It breaks on anything nobody anticipated (the scammer changes one word and gets through).

Rules

AI systemMachine learning

How it works: It's shown millions of examples (emails labeled spam and not spam) and learns the patterns on its own, including ones no person could have spelled out.
Strength: It adapts to new cases that look like the ones it has seen.
Limit: It can be wrong and can't always explain why it decided something.

Examples

The terms you'll keep hearing

Term	What it means in plain English
Artificial intelligence (AI)	The whole field: machines that perform "intelligent" tasks.
Machine learning (ML)	Teaching machines with data instead of hand-written rules.
Deep learning	Machine learning with many-layered neural networks. It's what made modern speech and image recognition possible.
Generative AI	AI that creates new content: text, images, audio, video or code.
Large language model (LLM)	A generative AI model trained on huge amounts of text. ChatGPT, Claude, Gemini and Copilot all run on LLMs.
Prompt	The instruction or question you type to the AI.
AI agent	An AI system that doesn't just answer but carries out multistep tasks using tools (searching the web, opening files, sending messages).

These words aren't synonyms. Each one sits inside the one before it:

Artificial intelligence the fieldAny system that performs "intelligent" tasks, including rule-based ones.
Machine and deep learning the techniqueLearning from data; deep learning uses many-layered neural networks.
Generative AI what it createsNew text, images, audio, video or code.
Large language models the engineWhat powers ChatGPT, Claude, Gemini and Copilot.

Every LLM is generative AI, but not all AI is generative: your bank's fraud detection is AI too.

1. What is generative AI?

AI that only sorts data into categories
AI that creates new content: text, images, audio or code
Physical robots that perform tasks

Generative AI produces new content based on patterns it learned. ChatGPT, Claude and Gemini are examples; a spam filter that labels email is AI, but not generative.

AI you were already using without knowing it

Long before chatbots, AI was part of your day. Most of it falls into three families:

It recommends

The shows Netflix suggests, the songs on your Spotify mixes and the "customers also bought" row on Amazon.

What it learns: what you did and what people like you did.

It protects

Your email's spam filter, the fraud alert from your credit card company and face unlock on your phone.

What it learns: what normal looks like, so it can flag what's odd.

It understands and predicts

Machine translation, the keyboard that guesses your next word, searching your photos for "dog" and the map app that routes you around traffic.

What it learns: patterns in language, images and traffic.

More and more companies also use it to serve you. Bank of America's virtual assistant, Erica, has handled more than 3 billion client interactions since it launched in 2018, according to the bank (August 2025). Chances are you've already talked to an AI without noticing.

All of these are "predictive" (or "discriminative") AI: they classify, recommend or forecast. What's new about generative AI is that it creates, and you use it by simply talking to it in plain language.

2. Which of these is an example of AI you probably use every day?

A pocket calculator
A PDF file
YouTube recommendations or your email's spam filter

Recommendations, spam filters, voice recognition and traffic-aware maps have used AI for years. A calculator follows fixed rules, and a PDF is just a file format.

The main generative AI tools

The market moves fast, but these are the categories and some of the best-known names (reviewed in November 2026):

Category	What it's for	Examples
Chat assistants	Writing, summarizing, explaining, analyzing, coding	ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google), Copilot (Microsoft), Meta AI, DeepSeek
AI inside the apps you use	AI in your documents, email and spreadsheets	Copilot in Microsoft 365, Gemini in Google Workspace, AI features in Notion, Canva or Zoom
Images	Creating or editing images from text	Image generators built into the assistants, Midjourney, Adobe Firefly, Canva
Audio and voice	Transcribing, dubbing, creating voices and music	Meeting transcription tools, ElevenLabs, Suno
Video	Generating or editing video	Runway, Veo, AI video editors
Coding	Writing and reviewing code	GitHub Copilot, Cursor, Claude Code, Codex

Don't get attached to one tool

Tools change every few months; skills don't. One example: OpenAI is retiring custom GPTs on December 11, 2026, less than three years after launching them, and replacing them with a new format (you'll see this in lesson 8). Writing good instructions, checking results and choosing which tasks to delegate work with any assistant. That's what this course teaches. Most assistants have a free version that's plenty for practice.

What generative AI does well (and what it doesn't)

Does well	Does badly or with risk
Drafting, rewriting and changing tone	Giving exact facts without a source (it may invent them)
Summarizing long texts and pulling out key points	Complex math without tools (though it's improving fast)
Generating ideas and alternatives	Very recent information if it has no web search
Explaining concepts at your level	Decisions that require your context, judgment or legal responsibility
Translating and adapting text	Final medical, legal or financial advice
Writing and explaining code and Excel formulas	Guaranteeing its output is correct without review

The golden rule for the whole course: AI is an extremely fast collaborator that is sometimes wrong with total confidence. You're still responsible for the result.

3. In which of these tasks should you double-check the AI's answer most carefully?

Exact figures, dates or quotes
Suggesting headline ideas
Rewriting an email in a friendlier tone

AI is excellent at drafting and brainstorming, but it can invent specific facts that sound completely believable. A weak headline idea costs nothing; a wrong number in a report can cost a lot.

Your first three tests (copy-and-paste prompts)

The best way to understand AI is to use it on something of your own. Copy these prompts, change what's in brackets and compare what you get with the expected result.

Test 1: have it explain something about your work

Explain generative AI as if I ran [a hardware store in Tulsa /
a dental office in Phoenix / a tutoring center in Atlanta]. Give
me 3 concrete ways I could use it this week without buying
anything. 150 words max.

Expected result: a short, jargon-free explanation and three uses you can apply right away (for example, answering common customer questions, writing product descriptions or drafting a social media post). If the ideas feel generic, add more context: your products, your customers, your biggest headache.

Test 2: a draft from your everyday work

Write a short, friendly message letting my customers know we'll
close at 2 p.m. on Saturday [date] for inventory. Warm tone,
50 words max, with an apology and an alternative (they can order
online or by text until Friday).

Expected result: a message ready to send with minor edits. Notice how much time it saved compared with writing it yourself.

Test 3: put it to the test with a fact

What is the current minimum wage in [your state], and since what
date has it applied? Name the official source, and if you're
not sure, tell me.

Expected result: a figure with its source, or a warning that it can't confirm it. Compare it with your state labor department's website or the U.S. Department of Labor. The federal minimum has been $7.25 an hour since 2009, but many states and cities set higher minimums that change every year, which is exactly where an assistant without web search tends to give stale numbers. If it doesn't match, you've just seen why verification is part of the method.

Common beginner mistakes

Asking something vague ("tell me about marketing") and concluding AI "doesn't work."

Give context, a goal and a format. You'll learn how in lesson 3.

Copying numbers, names or regulations without checking them.

Use AI for the draft and verify facts in the original source.

Pasting customers' personal data into a free tool.

Anonymize ("Customer A") or use only tools your company has approved.

Settling for the first answer.

Ask for specific changes: shorter, a different tone, three different versions.

Learning one tool as if it will last forever.

Learn the method: it works with any assistant, today's or tomorrow's.

4. Based on how the concepts in this lesson fit together, which statement is correct?

All artificial intelligence is generative AI.
A large language model is a type of generative AI, which in turn uses machine learning techniques.
Machine learning and large language model mean the same thing.

The concepts go from general to specific: AI ⊃ machine learning ⊃ deep learning ⊃ generative AI ⊃ LLMs. A credit card fraud detector, for example, is AI, but it isn't generative.

Why learn AI now

Workforce studies and employer surveys from the last few years agree on one thing: the tasks being automated fastest are writing, summarizing, looking things up and processing repetitive data. That doesn't necessarily mean jobs disappear, but it does mean they change. Someone who can use AI to do in one hour what used to take four has a clear edge in any role: sales, operations, teaching, marketing, legal, accounting or customer service.

And you don't have to spend money to start: the free versions of the major assistants are enough for the entire Beginner level of this course. What you do need is deliberate practice. Use it every day on a real task, compare the result with what you would have done and write down which prompts worked. Give it 15 minutes a day for two weeks and you'll notice the difference.

Practice exercise: your first conversation

Open any free assistant (ChatGPT, Claude, Gemini or Copilot) and try these three requests. Pay attention to how the answers differ:

"Explain inflation."
"Explain inflation as if I were 12, using an example from a neighborhood grocery store, in under 100 words."
"What was the U.S. inflation rate last month?"

See solution

The second answer is far more useful because you gave it context, an audience, a format and a length. That's the foundation of lesson 3.
On the third, if the assistant doesn't have web search, it may give you an old figure, tell you it doesn't know or, worse, make one up. Check the number against the official source: the Bureau of Labor Statistics publishes the Consumer Price Index (CPI) every month at bls.gov.
If the assistant did search the web, open the source it cites and confirm the month: "last month" is easy to get wrong when a new report has just come out.

Summary

AI learns patterns from data; generative AI creates new content.
Assistants like ChatGPT, Claude, Gemini and Copilot run on large language models (LLMs).
They're excellent at drafting, summarizing, explaining and brainstorming, and they can be wrong about facts.
Skills (giving good instructions and verifying) are worth more than any single tool.
Never paste personal or confidential data into a tool your company hasn't approved.

Next lesson: How AI Chatbots Work: Tokens, Context Windows and Hallucinations. You'll see how next-word prediction, training and the context window explain why AI sometimes invents facts, and what to do about it.