LEARN AI · CHAPTER 1 OF 9 · 5 MIN · FREE
WHAT AI IS
What tools like Claude actually are, what they are great at, and where they trip up.
BY THE END YOU CAN
- Explain in one sentence what Claude is and how it writes
- Name tasks it does well and tasks to keep away from it
- Spot why it can sound sure and still be wrong
5 short lessons · about 5 minutes · 4-question quiz. Prefer one idea per screen? Take the interactive chapter.
LESSON 1 · A VERY WELL-READ ASSISTANT
Claude is a computer program that has read an enormous amount of writing: books, articles, websites and more.
From all that reading it learned how language works. When you write to it, it writes back, one small piece at a time, each time choosing a piece that fits well with what came before.
The word suggestions on your phone keyboard, grown up. Instead of suggesting one word, it can write a whole email, a plan or a story that fits what you asked.
- YOU WRITE"Help me reply to this complaint politely."
- CLAUDEwrites a good reply, piece by piece
- YOU GETa draft you can edit and send
LESSON 2 · WHAT IT IS GREAT AT
Anything that is mostly about words, ideas or patterns. It is fast, never tired, and happy to try again.
Emails, posts, proposals, job ads
Long reports, meeting notes, email threads
A contract clause, a chart, a new topic
Names, ideas, angles, questions to ask
Shorter, friendlier, simpler, translated
Spot themes in feedback or numbers in a file
If you would happily hand the task to a smart new colleague with a good brief, Claude can probably help.
LESSON 3 · WHAT IT IS NOT
It is not a search engine like Google. A search engine finds pages and shows you links. Claude writes an answer from what it learned, unless you switch on its web search, in which case it looks things up and shows its sources.
It is not a calculator, a database or a person. It does not know your company, your customers or yesterday's meeting unless you tell it.
- Finds existing pages
- Shows links
- You read and decide
- Writes a new answer
- Explains in your words
- Can be wrong; check what matters
LESSON 4 · IT CAN BE CONFIDENTLY WRONG
Because it writes what sounds likely, it can produce something that sounds right but is not: a made-up number, a fake quote, a wrong date. People call this "making things up".
It does not lie on purpose. It simply has no built-in sense of "I am not sure" unless you invite it to say so.
A brilliant new colleague who has read everything but is too eager to please. Ask them for a number they do not know, and they may give you a believable guess.
Add: "If you are not sure, say so." It noticeably cuts made-up answers.
LESSON 5 · WHERE TO FIND IT AND HOW TO START
You can use Claude in a web browser at claude.ai, in the desktop app for Mac and Windows, and in the phone app. Sign in with the same account and your chats follow you from one device to another.
There is a free plan to try it; paid plans give you more use and extra features. At work, your company may give you a company account with its own rules. Use that one for work.
To start, type a request the way you would ask a colleague, and press send. You can always ask a follow-up question.
Your email: the same inbox on your laptop and your phone.
CHECK YOURSELF: 4 QUESTIONS
1. You ask Claude for the current price of a rival's product, with web search off. What should you expect?
- The exact price, since it has read most of the internet
- An answer that may be out of date, or simply made up
- A polite refusal, since prices change too often to say
SHOW ANSWER
An answer that may be out of date, or simply made up Without web search it can only use what it learned, which may be old, and it may fill the gap with a guess.
2. Which of these jobs suits Claude best?
- Adding up 3,000 sales figures to the exact cent
- Turning messy meeting notes into a clear summary
- Finding out what a client said on yesterday's call
SHOW ANSWER
Turning messy meeting notes into a clear summary Summarising text is a strength. Exact sums belong in a spreadsheet, and it cannot know about a call nobody told it about.
3. Which line in your message does most to cut made-up answers?
- "Please be 100% accurate and double-check yourself."
- "If you are not sure, say so instead of guessing."
- "You are a world-class expert in this subject."
SHOW ANSWER
"If you are not sure, say so instead of guessing." Giving it permission to admit doubt is one of Anthropic's own recommended ways to reduce made-up facts. Asking for accuracy does not give it a way out.
4. A colleague says "Claude already knows our refund policy." When can you rely on that?
- When the policy is published somewhere on our website
- When the policy is in the chat, a Project or a linked app
- When we use a paid plan instead of the free one
SHOW ANSWER
When the policy is in the chat, a Project or a linked app Even if it once read your website, it may remember it wrongly or out of date. It reliably knows your documents only when you give them to it.
GO DEEPER: FURTHER READING
Hand-picked guides, articles and videos on this chapter's topics. Links checked 2026-11-06.
- GUIDEGet started with Claude (Claude Help Center)
Anthropic's own first steps: what Claude is, where to use it, and tips for your first prompts.
- ARTICLEThe Anthropic Economic Index
Anthropic research on which real work tasks people give to Claude, and how that is changing.
- ARTICLEOne Useful Thing (Ethan Mollick)
A plain-English blog by a business school professor on using AI at work, with real examples.