Start Here › AI 101
AI 101: The Plain-English Basics
What AI is, what it isn’t, and how to start using it in five minutes.
Before we start: five things that are obvious, and often overlooked
- Building AI is not the same as using it.
- You don’t need to be an engineer to use it well.
- There are no experts in using AI. Everyone is learning as they go, including us.
- There is no official license or test. Anyone can call themselves an AI expert, sell AI consulting, or build and release an AI app.
- What you don’t know can hurt you.
The standard we hold ourselves to
Long ago we adopted what we call the Fiduciary Standard of Care for all of our work on deconstructingbabel.com. The reason is simple: the most consequential technology in history ought to be held to the highest standard in commerce.
A rung up: A fiduciary is someone bound to put your interests ahead of their own, the way a trustee or a fee-only financial adviser must. Holding ourselves to that standard means we owe you honesty first, even when it isn’t convenient for us.
In practice, that means no conflicts of interest, full transparency, and the duty of care you’d expect in any field where mistakes are costly. It doesn’t mean we’ll always be right. No one is right all the time. Trust, but verify. That’s why we include all those annoying footnotes: so you can check for yourself. We work hard not to make mistakes, and when we do, we correct them in the open. If you spot one, please tell us.
Having spent decades in consequential, highly regulated fields, we expect vigorous regulation of AI. In fact we’re pushing for it, and building for it.
Are we for AI or against it?
That’s the question we get most. Our cheeky answer: ask us whether we like gravity. It amounts to the same thing. Whether we like it doesn’t matter, and gravity isn’t always so great either. It is here, and we have to deal with it.
Over a year ago we wrote, “It’s here, it’s queer, get used to it.” Now that we have a new thing in our midst, it’s time to make friends. Welcome to the future.
What is AI, really?
“Artificial intelligence” is a broad term for computer programs that do things we used to think only people could do: understand language, recognize faces, translate, answer questions, write. It is neither artificial nor intelligent.
When most people say “AI” today, they mean a chatbot, also called an AI assistant. It’s a program you can talk to in ordinary language, by typing or out loud, and it answers in ordinary language. The best-known ones are ChatGPT, Claude, Gemini, Microsoft Copilot and Perplexity.
A rung up: Think of the world’s most well-read assistant. It has read a huge share of everything people have written: books, websites, articles. From all that reading it has learned how language fits together, so when you ask it something, it builds an answer one word at a time from the patterns it learned. It isn’t looking the answer up in a filing cabinet. It’s writing a fresh reply.
What AI is good at
- Explaining things: a confusing letter from your insurance company, a medical term, how a tax form works.
- Writing help: an email, a thank-you note, a complaint letter, a wedding toast.
- Summarizing: turning a long document into a few key points.
- Planning: a week of meals, a trip, a garden layout, a workout routine.
- “How do I…” questions: fixing a dripping faucet, setting up a new phone, using a spreadsheet.
- Brainstorming: gift ideas, names for a small business, things to do with the grandkids on a rainy day.
- Translating: menus, letters, conversations.
- Patient teaching: you can ask the same question five times in five ways, and it never gets tired of you.
What AI is not good at (and where it can mislead you)
- It can be confidently wrong. AI sometimes makes up facts, numbers, quotes or even sources that sound completely real. The industry calls this a “hallucination.” We call it making stuff up.
- It isn’t a doctor, lawyer or financial adviser. It can help you understand your situation and prepare good questions. It should not have the last word on your health, your money or your legal rights.
- It doesn’t know you unless you tell it. It can’t see your situation, your history or your values unless you explain them.
- It may not know about recent events unless the tool can search the web. Many tools now can, but check.
- It isn’t a person. It can sound warm, even caring. That’s useful, but remember it’s a tool, not a friend who will notice if you’re not okay.
A rung up: Why does AI make things up? Because it’s built to produce answers that sound right. Usually sounding right and being right go together, but not always. When it doesn’t know, it may fill the gap with something plausible instead of saying “I don’t know.” That’s why the golden rule below exists.
The golden rule: Use AI like a smart, fast, sometimes careless helper. Let it do the first draft and the legwork. You do the checking, and you make the final call.
How to get started in five minutes
- Pick one assistant. Most have a free version. Go to its website on your computer, or download its app from your phone’s app store. Get the app only from the official store, so you don’t download a fake.
- Create an account with your email. Use a strong password you don’t use anywhere else.
- Type a simple question, the way you’d ask a knowledgeable friend. For example: “What’s the best way to keep deer out of my vegetable garden in New Hampshire?”
- Read the answer, then keep talking. Ask, “Can you make that shorter?” or “Why does that work?” or “What would you do if you were me?”
- That’s it. You’re using AI.
Chatbots and agents: what’s the difference?
You’ll start hearing the word agent more and more. Here’s the plain difference.
- A chatbot talks. You ask, it answers. Nothing happens in the world unless you act on what it said.
- An agent acts. You give it a goal, and it takes steps for you: it can click through websites, fill in forms, send emails, book a table or buy something, depending on what you allow it to reach.
A rung up: A chatbot is like a knowledgeable friend on the phone. An agent is like handing that friend your car keys and your credit card. Both can be helpful. Only one can drive off with your money by mistake.
Three simple rules for agents:
- Give an agent only the access you’d give a new assistant on their first day.
- Make it ask you before it spends money, sends anything in your name, or deletes anything.
- Check its work afterward, the same way you would a person’s.
A few words you’ll hear
- Chatbot / AI assistant: the program you talk to.
- Agent: an AI that can take actions for you, not just answer.
- Prompt: whatever you type or say to the AI. Your question or request.
- Hallucination: when the AI states something false as if it were true.
- Large language model (LLM): the technical name for the engine inside a chatbot.
- Voice mode: talking to the AI out loud instead of typing.
- Deepfake: a fake photo, video or voice made with AI to look or sound like a real person.
Want more words? See our full Glossary.
Is it safe?
Mostly yes, if you use common sense. Treat anything you type into an AI the way you’d treat a postcard: assume someone else could read it. Don’t type Social Security numbers, bank account numbers, passwords or private medical details you wouldn’t want shared. Most tools let you turn off chat history or keep your chats out of the company’s training. Look in the settings under “Privacy” or “Data controls.”
The bigger safety issue isn’t the AI you use. It’s the AI criminals use against you. That’s covered on the next page.
Go deeper: read What Is AI—And Why You Should Care and our AI User Manual series, written for teachers, nurses, small-business owners, home managers and patients.
Keep going